init

README.md

@@ -0,0 +1,68 @@
+---
+license: other
+library_name: peft
+tags:
+- llama-factory
+- lora
+- generated_from_trainer
+base_model: mistralai/Mistral-7B-Instruct-v0.2
+model-index:
+- name: Mistral-7b_5000_8epoch_lora
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# Mistral-7b_5000_8epoch_lora
+
+This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on the advbench_sft_malicious dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.8734
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0001
+- train_batch_size: 2
+- eval_batch_size: 1
+- seed: 42
+- distributed_type: multi-GPU
+- num_devices: 4
+- gradient_accumulation_steps: 4
+- total_train_batch_size: 32
+- total_eval_batch_size: 4
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- num_epochs: 8.0
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| 0.3101        | 2.9542 | 500  | 0.4940          |
+| 0.0918        | 5.9084 | 1000 | 0.7312          |
+
+
+### Framework versions
+
+- PEFT 0.11.1
+- Transformers 4.41.2
+- Pytorch 2.3.1+cu121
+- Datasets 2.19.2
+- Tokenizers 0.19.1

adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "o_proj",
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e46a888997d55b5ee05c2a590a771639dd2fc91dd84b39ea2729268d8bccb278
+size 42002136

all_results.json

@@ -0,0 +1,12 @@
+{
+    "epoch": 7.988183161004431,
+    "eval_loss": 0.8734431862831116,
+    "eval_runtime": 52.5704,
+    "eval_samples_per_second": 2.111,
+    "eval_steps_per_second": 0.533,
+    "total_flos": 1435586865135616.0,
+    "train_loss": 0.22072081432606167,
+    "train_runtime": 22818.9262,
+    "train_samples_per_second": 1.896,
+    "train_steps_per_second": 0.059
+}

checkpoint-1000/README.md

@@ -0,0 +1,202 @@
+---
+library_name: peft
+base_model: mistralai/Mistral-7B-Instruct-v0.2
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.11.1

checkpoint-1000/adapter_config.json

@@ -0,0 +1,34 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "mistralai/Mistral-7B-Instruct-v0.2",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_dropout": 0.0,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "down_proj",
+    "o_proj",
+    "v_proj",
+    "q_proj"
+  ],
+  "task_type": "CAUSAL_LM",
+  "use_dora": false,
+  "use_rslora": false
+}

checkpoint-1000/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8158a0cefdf16f8ff8897fccfc3b82bbfa8bb515879844d32a8b2f923e5749d
+size 42002136

checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ebaaef666b7b2aeb07cdfbd69562c2265512c7a72cfa5852a5b5654da9f816c
+size 425070

checkpoint-1000/global_step1000/zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ed779a8897cda6945df3def88b7abd2a6f8ad6e7d56840ebb9916c58fc9ada3
+size 62918600

checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:136712c6e13dbae5bc5cd271fc43853555abc8ca876650697f0a0233957f8056
+size 425070

checkpoint-1000/global_step1000/zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cb9eb8e01a34204925e262b2828bc85f92ac3e69a471f94b4e91261f3528bf0
+size 62918600

checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:663a6d99f0b1148e533a3711a773de08c6825750fdf0117c9175a0d8b80b1507
+size 425070

checkpoint-1000/global_step1000/zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60a2dabb61276b0f2a8c43f4dfa40cb08e88d7bee39c8be744d6593991bc68df
+size 62918600

checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d43209cd9bd6f8c409515946bce50f6491a7b49f117e577b9b82f9d7f3122286
+size 425070

checkpoint-1000/global_step1000/zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae50ec80e67a0f92ff015d127b037df59c1062196b78264035a192dcfc1108e5
+size 62918600

checkpoint-1000/latest

@@ -0,0 +1 @@
+global_step1000

checkpoint-1000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee97cd82dba4d425fdd8dfdb88d4a43d0d4b1979b5c81ab4a24914fb00d4f332
+size 15024

checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91dad95440fb85dc4a31745642117165c1a72173b2e389679ea8c0b2b6fcd7e2
+size 15024

checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98698326b023c2af02c94f18726ce52c7f7a6fe290734dd7edbe99bc807fcfa0
+size 15024

checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:708e7c6b5bf8a327e688779ebc08830ce249928bcb1ff5c82b1b1d0bf6d2660b
+size 15024

checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:017e3823021c92614a8748b0ba13d32c7edd9b19c342e69c1c85e0c533ef0a7f
+size 1064

checkpoint-1000/special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

checkpoint-1000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

checkpoint-1000/tokenizer_config.json

@@ -0,0 +1,45 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{{ '<s>' + system_message }}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ ' [INST] ' + content + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "split_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-1000/trainer_state.json

@@ -0,0 +1,749 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.908419497784343,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.059084194977843424,
+      "grad_norm": 0.6387189059092414,
+      "learning_rate": 9.998650208062712e-05,
+      "loss": 0.6972,
+      "step": 10
+    },
+    {
+      "epoch": 0.11816838995568685,
+      "grad_norm": 0.48601036335500336,
+      "learning_rate": 9.994601561026155e-05,
+      "loss": 0.5423,
+      "step": 20
+    },
+    {
+      "epoch": 0.17725258493353027,
+      "grad_norm": 0.40437591619558344,
+      "learning_rate": 9.98785624482278e-05,
+      "loss": 0.5149,
+      "step": 30
+    },
+    {
+      "epoch": 0.2363367799113737,
+      "grad_norm": 0.5030211090458666,
+      "learning_rate": 9.978417901361958e-05,
+      "loss": 0.513,
+      "step": 40
+    },
+    {
+      "epoch": 0.29542097488921715,
+      "grad_norm": 0.49098083558596206,
+      "learning_rate": 9.96629162656365e-05,
+      "loss": 0.4957,
+      "step": 50
+    },
+    {
+      "epoch": 0.35450516986706054,
+      "grad_norm": 0.6329012343929873,
+      "learning_rate": 9.951483967607041e-05,
+      "loss": 0.495,
+      "step": 60
+    },
+    {
+      "epoch": 0.413589364844904,
+      "grad_norm": 0.48252149699962754,
+      "learning_rate": 9.934002919395592e-05,
+      "loss": 0.4943,
+      "step": 70
+    },
+    {
+      "epoch": 0.4726735598227474,
+      "grad_norm": 0.5221971068930835,
+      "learning_rate": 9.91385792024048e-05,
+      "loss": 0.4738,
+      "step": 80
+    },
+    {
+      "epoch": 0.5317577548005908,
+      "grad_norm": 0.42549092560536134,
+      "learning_rate": 9.891059846764679e-05,
+      "loss": 0.4565,
+      "step": 90
+    },
+    {
+      "epoch": 0.5908419497784343,
+      "grad_norm": 0.4606171719652433,
+      "learning_rate": 9.865621008030492e-05,
+      "loss": 0.4674,
+      "step": 100
+    },
+    {
+      "epoch": 0.6499261447562777,
+      "grad_norm": 0.46353540699946943,
+      "learning_rate": 9.83755513889369e-05,
+      "loss": 0.4727,
+      "step": 110
+    },
+    {
+      "epoch": 0.7090103397341211,
+      "grad_norm": 0.4815928480268326,
+      "learning_rate": 9.80687739258782e-05,
+      "loss": 0.4736,
+      "step": 120
+    },
+    {
+      "epoch": 0.7680945347119645,
+      "grad_norm": 0.485838392040906,
+      "learning_rate": 9.773604332542729e-05,
+      "loss": 0.47,
+      "step": 130
+    },
+    {
+      "epoch": 0.827178729689808,
+      "grad_norm": 0.5025850953484241,
+      "learning_rate": 9.737753923441688e-05,
+      "loss": 0.467,
+      "step": 140
+    },
+    {
+      "epoch": 0.8862629246676514,
+      "grad_norm": 0.45997226542102815,
+      "learning_rate": 9.69934552152196e-05,
+      "loss": 0.4522,
+      "step": 150
+    },
+    {
+      "epoch": 0.9453471196454948,
+      "grad_norm": 0.4546309634161405,
+      "learning_rate": 9.658399864124037e-05,
+      "loss": 0.4613,
+      "step": 160
+    },
+    {
+      "epoch": 1.0044313146233383,
+      "grad_norm": 0.6577670848472944,
+      "learning_rate": 9.61493905849521e-05,
+      "loss": 0.4641,
+      "step": 170
+    },
+    {
+      "epoch": 1.0635155096011817,
+      "grad_norm": 0.5743711541659764,
+      "learning_rate": 9.568986569853487e-05,
+      "loss": 0.3946,
+      "step": 180
+    },
+    {
+      "epoch": 1.122599704579025,
+      "grad_norm": 0.5374571463070855,
+      "learning_rate": 9.520567208718337e-05,
+      "loss": 0.3882,
+      "step": 190
+    },
+    {
+      "epoch": 1.1816838995568686,
+      "grad_norm": 0.6531134738909915,
+      "learning_rate": 9.469707117515067e-05,
+      "loss": 0.4205,
+      "step": 200
+    },
+    {
+      "epoch": 1.240768094534712,
+      "grad_norm": 0.5201345622237219,
+      "learning_rate": 9.416433756460091e-05,
+      "loss": 0.386,
+      "step": 210
+    },
+    {
+      "epoch": 1.2998522895125553,
+      "grad_norm": 0.7267530776250423,
+      "learning_rate": 9.360775888734698e-05,
+      "loss": 0.4096,
+      "step": 220
+    },
+    {
+      "epoch": 1.3589364844903988,
+      "grad_norm": 0.5480147887972682,
+      "learning_rate": 9.302763564955331e-05,
+      "loss": 0.3921,
+      "step": 230
+    },
+    {
+      "epoch": 1.4180206794682422,
+      "grad_norm": 0.6228502280992079,
+      "learning_rate": 9.242428106948749e-05,
+      "loss": 0.3788,
+      "step": 240
+    },
+    {
+      "epoch": 1.4771048744460857,
+      "grad_norm": 0.7119116813192373,
+      "learning_rate": 9.179802090840853e-05,
+      "loss": 0.3894,
+      "step": 250
+    },
+    {
+      "epoch": 1.536189069423929,
+      "grad_norm": 0.5987495739217404,
+      "learning_rate": 9.114919329468282e-05,
+      "loss": 0.3707,
+      "step": 260
+    },
+    {
+      "epoch": 1.5952732644017726,
+      "grad_norm": 0.6278387167291306,
+      "learning_rate": 9.04781485412231e-05,
+      "loss": 0.3938,
+      "step": 270
+    },
+    {
+      "epoch": 1.654357459379616,
+      "grad_norm": 0.5312391005336979,
+      "learning_rate": 8.978524895634842e-05,
+      "loss": 0.3799,
+      "step": 280
+    },
+    {
+      "epoch": 1.7134416543574593,
+      "grad_norm": 0.7329625126762797,
+      "learning_rate": 8.907086864816803e-05,
+      "loss": 0.403,
+      "step": 290
+    },
+    {
+      "epoch": 1.7725258493353029,
+      "grad_norm": 0.5600401734580108,
+      "learning_rate": 8.833539332259398e-05,
+      "loss": 0.3758,
+      "step": 300
+    },
+    {
+      "epoch": 1.8316100443131462,
+      "grad_norm": 0.503395394356135,
+      "learning_rate": 8.757922007509207e-05,
+      "loss": 0.3963,
+      "step": 310
+    },
+    {
+      "epoch": 1.8906942392909896,
+      "grad_norm": 0.6324775953771359,
+      "learning_rate": 8.680275717628337e-05,
+      "loss": 0.3858,
+      "step": 320
+    },
+    {
+      "epoch": 1.9497784342688331,
+      "grad_norm": 0.5574107174736728,
+      "learning_rate": 8.600642385151205e-05,
+      "loss": 0.3799,
+      "step": 330
+    },
+    {
+      "epoch": 2.0088626292466767,
+      "grad_norm": 0.5250864362886176,
+      "learning_rate": 8.519065005449858e-05,
+      "loss": 0.3763,
+      "step": 340
+    },
+    {
+      "epoch": 2.06794682422452,
+      "grad_norm": 0.7754155552722259,
+      "learning_rate": 8.43558762352005e-05,
+      "loss": 0.2934,
+      "step": 350
+    },
+    {
+      "epoch": 2.1270310192023634,
+      "grad_norm": 0.6889925166837503,
+      "learning_rate": 8.350255310200612e-05,
+      "loss": 0.3078,
+      "step": 360
+    },
+    {
+      "epoch": 2.186115214180207,
+      "grad_norm": 0.6281519102970855,
+      "learning_rate": 8.263114137838947e-05,
+      "loss": 0.3028,
+      "step": 370
+    },
+    {
+      "epoch": 2.24519940915805,
+      "grad_norm": 0.7619069663070173,
+      "learning_rate": 8.174211155415799e-05,
+      "loss": 0.2972,
+      "step": 380
+    },
+    {
+      "epoch": 2.3042836041358936,
+      "grad_norm": 0.6469214076306162,
+      "learning_rate": 8.083594363142717e-05,
+      "loss": 0.2995,
+      "step": 390
+    },
+    {
+      "epoch": 2.363367799113737,
+      "grad_norm": 0.7761443706661119,
+      "learning_rate": 7.991312686545937e-05,
+      "loss": 0.2963,
+      "step": 400
+    },
+    {
+      "epoch": 2.4224519940915803,
+      "grad_norm": 0.6652219653359559,
+      "learning_rate": 7.897415950050676e-05,
+      "loss": 0.2987,
+      "step": 410
+    },
+    {
+      "epoch": 2.481536189069424,
+      "grad_norm": 0.7586700003856579,
+      "learning_rate": 7.801954850080075e-05,
+      "loss": 0.3092,
+      "step": 420
+    },
+    {
+      "epoch": 2.5406203840472674,
+      "grad_norm": 0.6663785963305503,
+      "learning_rate": 7.704980927683359e-05,
+      "loss": 0.2951,
+      "step": 430
+    },
+    {
+      "epoch": 2.5997045790251105,
+      "grad_norm": 0.6913254608481674,
+      "learning_rate": 7.60654654070796e-05,
+      "loss": 0.3097,
+      "step": 440
+    },
+    {
+      "epoch": 2.658788774002954,
+      "grad_norm": 0.6943652391419598,
+      "learning_rate": 7.506704835530634e-05,
+      "loss": 0.2999,
+      "step": 450
+    },
+    {
+      "epoch": 2.7178729689807977,
+      "grad_norm": 0.644096596125316,
+      "learning_rate": 7.405509718362842e-05,
+      "loss": 0.2905,
+      "step": 460
+    },
+    {
+      "epoch": 2.7769571639586412,
+      "grad_norm": 0.7660517390503399,
+      "learning_rate": 7.303015826145885e-05,
+      "loss": 0.309,
+      "step": 470
+    },
+    {
+      "epoch": 2.8360413589364843,
+      "grad_norm": 0.7708918433168639,
+      "learning_rate": 7.199278497051498e-05,
+      "loss": 0.302,
+      "step": 480
+    },
+    {
+      "epoch": 2.895125553914328,
+      "grad_norm": 0.626692120748867,
+      "learning_rate": 7.094353740603839e-05,
+      "loss": 0.297,
+      "step": 490
+    },
+    {
+      "epoch": 2.9542097488921715,
+      "grad_norm": 0.7926892798861292,
+      "learning_rate": 6.988298207439021e-05,
+      "loss": 0.3101,
+      "step": 500
+    },
+    {
+      "epoch": 2.9542097488921715,
+      "eval_loss": 0.4939613938331604,
+      "eval_runtime": 53.5552,
+      "eval_samples_per_second": 2.073,
+      "eval_steps_per_second": 0.523,
+      "step": 500
+    },
+    {
+      "epoch": 3.0132939438700146,
+      "grad_norm": 0.5747285632843896,
+      "learning_rate": 6.881169158718474e-05,
+      "loss": 0.2736,
+      "step": 510
+    },
+    {
+      "epoch": 3.072378138847858,
+      "grad_norm": 0.7661842329328333,
+      "learning_rate": 6.773024435212678e-05,
+      "loss": 0.2187,
+      "step": 520
+    },
+    {
+      "epoch": 3.1314623338257017,
+      "grad_norm": 0.8401289392682185,
+      "learning_rate": 6.663922426071977e-05,
+      "loss": 0.2057,
+      "step": 530
+    },
+    {
+      "epoch": 3.1905465288035453,
+      "grad_norm": 0.7506650372127406,
+      "learning_rate": 6.553922037301283e-05,
+      "loss": 0.2067,
+      "step": 540
+    },
+    {
+      "epoch": 3.2496307237813884,
+      "grad_norm": 0.7842941173009434,
+      "learning_rate": 6.443082659955738e-05,
+      "loss": 0.1989,
+      "step": 550
+    },
+    {
+      "epoch": 3.308714918759232,
+      "grad_norm": 0.7593744868915973,
+      "learning_rate": 6.331464138074493e-05,
+      "loss": 0.2179,
+      "step": 560
+    },
+    {
+      "epoch": 3.3677991137370755,
+      "grad_norm": 0.756091123285611,
+      "learning_rate": 6.219126736369903e-05,
+      "loss": 0.2176,
+      "step": 570
+    },
+    {
+      "epoch": 3.4268833087149186,
+      "grad_norm": 0.7869512689224245,
+      "learning_rate": 6.106131107689599e-05,
+      "loss": 0.2215,
+      "step": 580
+    },
+    {
+      "epoch": 3.485967503692762,
+      "grad_norm": 0.7636724676762149,
+      "learning_rate": 5.9925382602689974e-05,
+      "loss": 0.2153,
+      "step": 590
+    },
+    {
+      "epoch": 3.5450516986706058,
+      "grad_norm": 0.6815318566179079,
+      "learning_rate": 5.8784095247919305e-05,
+      "loss": 0.2133,
+      "step": 600
+    },
+    {
+      "epoch": 3.604135893648449,
+      "grad_norm": 0.7848948829508617,
+      "learning_rate": 5.763806521277184e-05,
+      "loss": 0.2109,
+      "step": 610
+    },
+    {
+      "epoch": 3.6632200886262924,
+      "grad_norm": 0.7715200213470335,
+      "learning_rate": 5.648791125808809e-05,
+      "loss": 0.2214,
+      "step": 620
+    },
+    {
+      "epoch": 3.722304283604136,
+      "grad_norm": 0.6480973448749833,
+      "learning_rate": 5.5334254371281934e-05,
+      "loss": 0.212,
+      "step": 630
+    },
+    {
+      "epoch": 3.781388478581979,
+      "grad_norm": 0.6618796234383102,
+      "learning_rate": 5.417771743105907e-05,
+      "loss": 0.2196,
+      "step": 640
+    },
+    {
+      "epoch": 3.8404726735598227,
+      "grad_norm": 0.9462710180746308,
+      "learning_rate": 5.3018924871114305e-05,
+      "loss": 0.2145,
+      "step": 650
+    },
+    {
+      "epoch": 3.8995568685376663,
+      "grad_norm": 0.7674586984793125,
+      "learning_rate": 5.185850234298942e-05,
+      "loss": 0.2199,
+      "step": 660
+    },
+    {
+      "epoch": 3.9586410635155094,
+      "grad_norm": 0.8420752213128357,
+      "learning_rate": 5.0697076378273354e-05,
+      "loss": 0.218,
+      "step": 670
+    },
+    {
+      "epoch": 4.017725258493353,
+      "grad_norm": 0.6235542870137168,
+      "learning_rate": 4.953527405032723e-05,
+      "loss": 0.1987,
+      "step": 680
+    },
+    {
+      "epoch": 4.0768094534711965,
+      "grad_norm": 0.8247620064480204,
+      "learning_rate": 4.8373722635717086e-05,
+      "loss": 0.1425,
+      "step": 690
+    },
+    {
+      "epoch": 4.13589364844904,
+      "grad_norm": 0.806070554149328,
+      "learning_rate": 4.721304927553658e-05,
+      "loss": 0.1313,
+      "step": 700
+    },
+    {
+      "epoch": 4.194977843426884,
+      "grad_norm": 1.1960115947293068,
+      "learning_rate": 4.60538806368031e-05,
+      "loss": 0.1397,
+      "step": 710
+    },
+    {
+      "epoch": 4.254062038404727,
+      "grad_norm": 0.7571134276917469,
+      "learning_rate": 4.489684257410958e-05,
+      "loss": 0.1421,
+      "step": 720
+    },
+    {
+      "epoch": 4.31314623338257,
+      "grad_norm": 0.8865025891479803,
+      "learning_rate": 4.374255979171538e-05,
+      "loss": 0.1386,
+      "step": 730
+    },
+    {
+      "epoch": 4.372230428360414,
+      "grad_norm": 0.9290220213968287,
+      "learning_rate": 4.2591655506257645e-05,
+      "loss": 0.1444,
+      "step": 740
+    },
+    {
+      "epoch": 4.431314623338257,
+      "grad_norm": 0.8896418597603776,
+      "learning_rate": 4.144475111026643e-05,
+      "loss": 0.1391,
+      "step": 750
+    },
+    {
+      "epoch": 4.4903988183161,
+      "grad_norm": 0.9053211649213782,
+      "learning_rate": 4.030246583666437e-05,
+      "loss": 0.1438,
+      "step": 760
+    },
+    {
+      "epoch": 4.549483013293944,
+      "grad_norm": 0.8453862385052026,
+      "learning_rate": 3.9165416424432414e-05,
+      "loss": 0.1415,
+      "step": 770
+    },
+    {
+      "epoch": 4.608567208271787,
+      "grad_norm": 0.8724405655899441,
+      "learning_rate": 3.803421678562213e-05,
+      "loss": 0.1492,
+      "step": 780
+    },
+    {
+      "epoch": 4.66765140324963,
+      "grad_norm": 0.98302153579186,
+      "learning_rate": 3.690947767389426e-05,
+      "loss": 0.1512,
+      "step": 790
+    },
+    {
+      "epoch": 4.726735598227474,
+      "grad_norm": 0.8137008256198869,
+      "learning_rate": 3.57918063547627e-05,
+      "loss": 0.1481,
+      "step": 800
+    },
+    {
+      "epoch": 4.7858197932053175,
+      "grad_norm": 0.8604546990181478,
+      "learning_rate": 3.468180627772144e-05,
+      "loss": 0.1418,
+      "step": 810
+    },
+    {
+      "epoch": 4.844903988183161,
+      "grad_norm": 0.8028241587093687,
+      "learning_rate": 3.358007675043224e-05,
+      "loss": 0.146,
+      "step": 820
+    },
+    {
+      "epoch": 4.903988183161005,
+      "grad_norm": 0.8209334713352179,
+      "learning_rate": 3.2487212615148316e-05,
+      "loss": 0.1407,
+      "step": 830
+    },
+    {
+      "epoch": 4.963072378138848,
+      "grad_norm": 0.9318035909918932,
+      "learning_rate": 3.1403803927549006e-05,
+      "loss": 0.1502,
+      "step": 840
+    },
+    {
+      "epoch": 5.022156573116692,
+      "grad_norm": 0.6982555180872486,
+      "learning_rate": 3.0330435638158806e-05,
+      "loss": 0.1322,
+      "step": 850
+    },
+    {
+      "epoch": 5.081240768094535,
+      "grad_norm": 0.9099529668785485,
+      "learning_rate": 2.9267687276522876e-05,
+      "loss": 0.0985,
+      "step": 860
+    },
+    {
+      "epoch": 5.140324963072378,
+      "grad_norm": 0.8093436219539469,
+      "learning_rate": 2.821613263830912e-05,
+      "loss": 0.0929,
+      "step": 870
+    },
+    {
+      "epoch": 5.199409158050222,
+      "grad_norm": 0.908675419764728,
+      "learning_rate": 2.717633947550651e-05,
+      "loss": 0.0941,
+      "step": 880
+    },
+    {
+      "epoch": 5.258493353028065,
+      "grad_norm": 0.9144782506889362,
+      "learning_rate": 2.614886918988604e-05,
+      "loss": 0.0951,
+      "step": 890
+    },
+    {
+      "epoch": 5.317577548005908,
+      "grad_norm": 0.8302439773379418,
+      "learning_rate": 2.5134276529890644e-05,
+      "loss": 0.0926,
+      "step": 900
+    },
+    {
+      "epoch": 5.376661742983752,
+      "grad_norm": 0.8145999759475107,
+      "learning_rate": 2.4133109291117156e-05,
+      "loss": 0.095,
+      "step": 910
+    },
+    {
+      "epoch": 5.435745937961595,
+      "grad_norm": 0.7459552675334057,
+      "learning_rate": 2.314590802055232e-05,
+      "loss": 0.0886,
+      "step": 920
+    },
+    {
+      "epoch": 5.4948301329394384,
+      "grad_norm": 0.9033419203896008,
+      "learning_rate": 2.2173205724722318e-05,
+      "loss": 0.096,
+      "step": 930
+    },
+    {
+      "epoch": 5.5539143279172825,
+      "grad_norm": 0.882377183289936,
+      "learning_rate": 2.121552758191366e-05,
+      "loss": 0.0962,
+      "step": 940
+    },
+    {
+      "epoch": 5.612998522895126,
+      "grad_norm": 0.8033415953079253,
+      "learning_rate": 2.027339065862064e-05,
+      "loss": 0.0985,
+      "step": 950
+    },
+    {
+      "epoch": 5.672082717872969,
+      "grad_norm": 0.9269892443045443,
+      "learning_rate": 1.934730363037237e-05,
+      "loss": 0.0939,
+      "step": 960
+    },
+    {
+      "epoch": 5.731166912850813,
+      "grad_norm": 0.8483635995759108,
+      "learning_rate": 1.843776650709046e-05,
+      "loss": 0.0975,
+      "step": 970
+    },
+    {
+      "epoch": 5.790251107828656,
+      "grad_norm": 0.7664349269568396,
+      "learning_rate": 1.7545270363125153e-05,
+      "loss": 0.093,
+      "step": 980
+    },
+    {
+      "epoch": 5.849335302806499,
+      "grad_norm": 0.7270945706246518,
+      "learning_rate": 1.6670297072116165e-05,
+      "loss": 0.0959,
+      "step": 990
+    },
+    {
+      "epoch": 5.908419497784343,
+      "grad_norm": 0.4763779090409354,
+      "learning_rate": 1.581331904682089e-05,
+      "loss": 0.0918,
+      "step": 1000
+    },
+    {
+      "epoch": 5.908419497784343,
+      "eval_loss": 0.7311862707138062,
+      "eval_runtime": 52.641,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 0.532,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1352,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1062799032713216.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:899f0486eced747a2f123ddafb9d6fecbc5fefa74f8054cc454aa966f7165bc2
+size 6968

checkpoint-1000/zero_to_fp32.py

@@ -0,0 +1,604 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 1, 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage <= 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dict = torch.load(f, map_location=device)
+        # immediately discard the potentially huge 2 optimizer states as we only care for fp32 master weights
+        # and also handle the case where it was already removed by another helper script
+        state_dict["optimizer_state_dict"].pop("optimizer_state_dict", None)
+        state_dicts.append(state_dict)
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage <= 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage <= 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage <= 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                                          exclude_frozen_parameters)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _has_callable(obj, fn):
+    attr = getattr(obj, fn, None)
+    return callable(attr)
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel() if _has_callable(shape, 'numel') else math.prod(shape)
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states,
+                                               exclude_frozen_parameters):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    if not exclude_frozen_parameters:
+        _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir, exclude_frozen_parameters)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None, exclude_frozen_parameters=False):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+        - ``exclude_frozen_parameters``: exclude frozen parameters
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag, exclude_frozen_parameters)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-t",
+                        "--tag",
+                        type=str,
+                        default=None,
+                        help="checkpoint tag used as a unique identifier for checkpoint. e.g., global_step1")
+    parser.add_argument("--exclude_frozen_parameters", action='store_true', help="exclude frozen parameters")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir,
+                                               args.output_file,
+                                               tag=args.tag,
+                                               exclude_frozen_parameters=args.exclude_frozen_parameters)

eval_results.json

@@ -0,0 +1,7 @@
+{
+    "epoch": 7.988183161004431,
+    "eval_loss": 0.8734431862831116,
+    "eval_runtime": 52.5704,
+    "eval_samples_per_second": 2.111,
+    "eval_steps_per_second": 0.533
+}

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "</s>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

tokenizer_config.json

@@ -0,0 +1,45 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{{ '<s>' + system_message }}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ ' [INST] ' + content + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' }}{% endif %}{% endfor %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "split_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 7.988183161004431,
+    "total_flos": 1435586865135616.0,
+    "train_loss": 0.22072081432606167,
+    "train_runtime": 22818.9262,
+    "train_samples_per_second": 1.896,
+    "train_steps_per_second": 0.059
+}

trainer_log.jsonl

@@ -0,0 +1,138 @@
+{"current_steps": 10, "total_steps": 1352, "loss": 0.6972, "learning_rate": 9.998650208062712e-05, "epoch": 0.059084194977843424, "percentage": 0.74, "elapsed_time": "0:02:55", "remaining_time": "6:31:38"}
+{"current_steps": 20, "total_steps": 1352, "loss": 0.5423, "learning_rate": 9.994601561026155e-05, "epoch": 0.11816838995568685, "percentage": 1.48, "elapsed_time": "0:05:43", "remaining_time": "6:21:30"}
+{"current_steps": 30, "total_steps": 1352, "loss": 0.5149, "learning_rate": 9.98785624482278e-05, "epoch": 0.17725258493353027, "percentage": 2.22, "elapsed_time": "0:08:36", "remaining_time": "6:18:58"}
+{"current_steps": 40, "total_steps": 1352, "loss": 0.513, "learning_rate": 9.978417901361958e-05, "epoch": 0.2363367799113737, "percentage": 2.96, "elapsed_time": "0:11:26", "remaining_time": "6:15:08"}
+{"current_steps": 50, "total_steps": 1352, "loss": 0.4957, "learning_rate": 9.96629162656365e-05, "epoch": 0.29542097488921715, "percentage": 3.7, "elapsed_time": "0:14:14", "remaining_time": "6:10:46"}
+{"current_steps": 60, "total_steps": 1352, "loss": 0.495, "learning_rate": 9.951483967607041e-05, "epoch": 0.35450516986706054, "percentage": 4.44, "elapsed_time": "0:17:01", "remaining_time": "6:06:31"}
+{"current_steps": 70, "total_steps": 1352, "loss": 0.4943, "learning_rate": 9.934002919395592e-05, "epoch": 0.413589364844904, "percentage": 5.18, "elapsed_time": "0:19:47", "remaining_time": "6:02:28"}
+{"current_steps": 80, "total_steps": 1352, "loss": 0.4738, "learning_rate": 9.91385792024048e-05, "epoch": 0.4726735598227474, "percentage": 5.92, "elapsed_time": "0:22:37", "remaining_time": "5:59:41"}
+{"current_steps": 90, "total_steps": 1352, "loss": 0.4565, "learning_rate": 9.891059846764679e-05, "epoch": 0.5317577548005908, "percentage": 6.66, "elapsed_time": "0:25:27", "remaining_time": "5:56:54"}
+{"current_steps": 100, "total_steps": 1352, "loss": 0.4674, "learning_rate": 9.865621008030492e-05, "epoch": 0.5908419497784343, "percentage": 7.4, "elapsed_time": "0:28:13", "remaining_time": "5:53:22"}
+{"current_steps": 110, "total_steps": 1352, "loss": 0.4727, "learning_rate": 9.83755513889369e-05, "epoch": 0.6499261447562777, "percentage": 8.14, "elapsed_time": "0:31:02", "remaining_time": "5:50:31"}
+{"current_steps": 120, "total_steps": 1352, "loss": 0.4736, "learning_rate": 9.80687739258782e-05, "epoch": 0.7090103397341211, "percentage": 8.88, "elapsed_time": "0:33:49", "remaining_time": "5:47:18"}
+{"current_steps": 130, "total_steps": 1352, "loss": 0.47, "learning_rate": 9.773604332542729e-05, "epoch": 0.7680945347119645, "percentage": 9.62, "elapsed_time": "0:36:36", "remaining_time": "5:44:02"}
+{"current_steps": 140, "total_steps": 1352, "loss": 0.467, "learning_rate": 9.737753923441688e-05, "epoch": 0.827178729689808, "percentage": 10.36, "elapsed_time": "0:39:26", "remaining_time": "5:41:26"}
+{"current_steps": 150, "total_steps": 1352, "loss": 0.4522, "learning_rate": 9.69934552152196e-05, "epoch": 0.8862629246676514, "percentage": 11.09, "elapsed_time": "0:42:15", "remaining_time": "5:38:37"}
+{"current_steps": 160, "total_steps": 1352, "loss": 0.4613, "learning_rate": 9.658399864124037e-05, "epoch": 0.9453471196454948, "percentage": 11.83, "elapsed_time": "0:45:01", "remaining_time": "5:35:24"}
+{"current_steps": 170, "total_steps": 1352, "loss": 0.4641, "learning_rate": 9.61493905849521e-05, "epoch": 1.0044313146233383, "percentage": 12.57, "elapsed_time": "0:47:48", "remaining_time": "5:32:25"}
+{"current_steps": 180, "total_steps": 1352, "loss": 0.3946, "learning_rate": 9.568986569853487e-05, "epoch": 1.0635155096011817, "percentage": 13.31, "elapsed_time": "0:50:42", "remaining_time": "5:30:07"}
+{"current_steps": 190, "total_steps": 1352, "loss": 0.3882, "learning_rate": 9.520567208718337e-05, "epoch": 1.122599704579025, "percentage": 14.05, "elapsed_time": "0:53:31", "remaining_time": "5:27:20"}
+{"current_steps": 200, "total_steps": 1352, "loss": 0.4205, "learning_rate": 9.469707117515067e-05, "epoch": 1.1816838995568686, "percentage": 14.79, "elapsed_time": "0:56:16", "remaining_time": "5:24:07"}
+{"current_steps": 210, "total_steps": 1352, "loss": 0.386, "learning_rate": 9.416433756460091e-05, "epoch": 1.240768094534712, "percentage": 15.53, "elapsed_time": "0:59:03", "remaining_time": "5:21:09"}
+{"current_steps": 220, "total_steps": 1352, "loss": 0.4096, "learning_rate": 9.360775888734698e-05, "epoch": 1.2998522895125553, "percentage": 16.27, "elapsed_time": "1:01:50", "remaining_time": "5:18:10"}
+{"current_steps": 230, "total_steps": 1352, "loss": 0.3921, "learning_rate": 9.302763564955331e-05, "epoch": 1.3589364844903988, "percentage": 17.01, "elapsed_time": "1:04:39", "remaining_time": "5:15:23"}
+{"current_steps": 240, "total_steps": 1352, "loss": 0.3788, "learning_rate": 9.242428106948749e-05, "epoch": 1.4180206794682422, "percentage": 17.75, "elapsed_time": "1:07:30", "remaining_time": "5:12:45"}
+{"current_steps": 250, "total_steps": 1352, "loss": 0.3894, "learning_rate": 9.179802090840853e-05, "epoch": 1.4771048744460857, "percentage": 18.49, "elapsed_time": "1:10:19", "remaining_time": "5:09:59"}
+{"current_steps": 260, "total_steps": 1352, "loss": 0.3707, "learning_rate": 9.114919329468282e-05, "epoch": 1.536189069423929, "percentage": 19.23, "elapsed_time": "1:13:12", "remaining_time": "5:07:27"}
+{"current_steps": 270, "total_steps": 1352, "loss": 0.3938, "learning_rate": 9.04781485412231e-05, "epoch": 1.5952732644017726, "percentage": 19.97, "elapsed_time": "1:15:55", "remaining_time": "5:04:16"}
+{"current_steps": 280, "total_steps": 1352, "loss": 0.3799, "learning_rate": 8.978524895634842e-05, "epoch": 1.654357459379616, "percentage": 20.71, "elapsed_time": "1:18:46", "remaining_time": "5:01:35"}
+{"current_steps": 290, "total_steps": 1352, "loss": 0.403, "learning_rate": 8.907086864816803e-05, "epoch": 1.7134416543574593, "percentage": 21.45, "elapsed_time": "1:21:33", "remaining_time": "4:58:40"}
+{"current_steps": 300, "total_steps": 1352, "loss": 0.3758, "learning_rate": 8.833539332259398e-05, "epoch": 1.7725258493353029, "percentage": 22.19, "elapsed_time": "1:24:28", "remaining_time": "4:56:12"}
+{"current_steps": 310, "total_steps": 1352, "loss": 0.3963, "learning_rate": 8.757922007509207e-05, "epoch": 1.8316100443131462, "percentage": 22.93, "elapsed_time": "1:27:16", "remaining_time": "4:53:23"}
+{"current_steps": 320, "total_steps": 1352, "loss": 0.3858, "learning_rate": 8.680275717628337e-05, "epoch": 1.8906942392909896, "percentage": 23.67, "elapsed_time": "1:30:06", "remaining_time": "4:50:35"}
+{"current_steps": 330, "total_steps": 1352, "loss": 0.3799, "learning_rate": 8.600642385151205e-05, "epoch": 1.9497784342688331, "percentage": 24.41, "elapsed_time": "1:32:52", "remaining_time": "4:47:37"}
+{"current_steps": 340, "total_steps": 1352, "loss": 0.3763, "learning_rate": 8.519065005449858e-05, "epoch": 2.0088626292466767, "percentage": 25.15, "elapsed_time": "1:35:37", "remaining_time": "4:44:37"}
+{"current_steps": 350, "total_steps": 1352, "loss": 0.2934, "learning_rate": 8.43558762352005e-05, "epoch": 2.06794682422452, "percentage": 25.89, "elapsed_time": "1:38:32", "remaining_time": "4:42:05"}
+{"current_steps": 360, "total_steps": 1352, "loss": 0.3078, "learning_rate": 8.350255310200612e-05, "epoch": 2.1270310192023634, "percentage": 26.63, "elapsed_time": "1:41:15", "remaining_time": "4:39:01"}
+{"current_steps": 370, "total_steps": 1352, "loss": 0.3028, "learning_rate": 8.263114137838947e-05, "epoch": 2.186115214180207, "percentage": 27.37, "elapsed_time": "1:44:01", "remaining_time": "4:36:06"}
+{"current_steps": 380, "total_steps": 1352, "loss": 0.2972, "learning_rate": 8.174211155415799e-05, "epoch": 2.24519940915805, "percentage": 28.11, "elapsed_time": "1:46:47", "remaining_time": "4:33:09"}
+{"current_steps": 390, "total_steps": 1352, "loss": 0.2995, "learning_rate": 8.083594363142717e-05, "epoch": 2.3042836041358936, "percentage": 28.85, "elapsed_time": "1:49:35", "remaining_time": "4:30:18"}
+{"current_steps": 400, "total_steps": 1352, "loss": 0.2963, "learning_rate": 7.991312686545937e-05, "epoch": 2.363367799113737, "percentage": 29.59, "elapsed_time": "1:52:24", "remaining_time": "4:27:32"}
+{"current_steps": 410, "total_steps": 1352, "loss": 0.2987, "learning_rate": 7.897415950050676e-05, "epoch": 2.4224519940915803, "percentage": 30.33, "elapsed_time": "1:55:12", "remaining_time": "4:24:40"}
+{"current_steps": 420, "total_steps": 1352, "loss": 0.3092, "learning_rate": 7.801954850080075e-05, "epoch": 2.481536189069424, "percentage": 31.07, "elapsed_time": "1:57:55", "remaining_time": "4:21:41"}
+{"current_steps": 430, "total_steps": 1352, "loss": 0.2951, "learning_rate": 7.704980927683359e-05, "epoch": 2.5406203840472674, "percentage": 31.8, "elapsed_time": "2:00:50", "remaining_time": "4:19:05"}
+{"current_steps": 440, "total_steps": 1352, "loss": 0.3097, "learning_rate": 7.60654654070796e-05, "epoch": 2.5997045790251105, "percentage": 32.54, "elapsed_time": "2:03:35", "remaining_time": "4:16:10"}
+{"current_steps": 450, "total_steps": 1352, "loss": 0.2999, "learning_rate": 7.506704835530634e-05, "epoch": 2.658788774002954, "percentage": 33.28, "elapsed_time": "2:06:27", "remaining_time": "4:13:28"}
+{"current_steps": 460, "total_steps": 1352, "loss": 0.2905, "learning_rate": 7.405509718362842e-05, "epoch": 2.7178729689807977, "percentage": 34.02, "elapsed_time": "2:09:15", "remaining_time": "4:10:37"}
+{"current_steps": 470, "total_steps": 1352, "loss": 0.309, "learning_rate": 7.303015826145885e-05, "epoch": 2.7769571639586412, "percentage": 34.76, "elapsed_time": "2:12:01", "remaining_time": "4:07:45"}
+{"current_steps": 480, "total_steps": 1352, "loss": 0.302, "learning_rate": 7.199278497051498e-05, "epoch": 2.8360413589364843, "percentage": 35.5, "elapsed_time": "2:14:51", "remaining_time": "4:05:00"}
+{"current_steps": 490, "total_steps": 1352, "loss": 0.297, "learning_rate": 7.094353740603839e-05, "epoch": 2.895125553914328, "percentage": 36.24, "elapsed_time": "2:17:38", "remaining_time": "4:02:07"}
+{"current_steps": 500, "total_steps": 1352, "loss": 0.3101, "learning_rate": 6.988298207439021e-05, "epoch": 2.9542097488921715, "percentage": 36.98, "elapsed_time": "2:20:26", "remaining_time": "3:59:18"}
+{"current_steps": 500, "total_steps": 1352, "eval_loss": 0.4939613938331604, "epoch": 2.9542097488921715, "percentage": 36.98, "elapsed_time": "2:21:20", "remaining_time": "4:00:49"}
+{"current_steps": 510, "total_steps": 1352, "loss": 0.2736, "learning_rate": 6.881169158718474e-05, "epoch": 3.0132939438700146, "percentage": 37.72, "elapsed_time": "2:24:11", "remaining_time": "3:58:03"}
+{"current_steps": 520, "total_steps": 1352, "loss": 0.2187, "learning_rate": 6.773024435212678e-05, "epoch": 3.072378138847858, "percentage": 38.46, "elapsed_time": "2:26:58", "remaining_time": "3:55:10"}
+{"current_steps": 530, "total_steps": 1352, "loss": 0.2057, "learning_rate": 6.663922426071977e-05, "epoch": 3.1314623338257017, "percentage": 39.2, "elapsed_time": "2:29:45", "remaining_time": "3:52:15"}
+{"current_steps": 540, "total_steps": 1352, "loss": 0.2067, "learning_rate": 6.553922037301283e-05, "epoch": 3.1905465288035453, "percentage": 39.94, "elapsed_time": "2:32:39", "remaining_time": "3:49:32"}
+{"current_steps": 550, "total_steps": 1352, "loss": 0.1989, "learning_rate": 6.443082659955738e-05, "epoch": 3.2496307237813884, "percentage": 40.68, "elapsed_time": "2:35:33", "remaining_time": "3:46:49"}
+{"current_steps": 560, "total_steps": 1352, "loss": 0.2179, "learning_rate": 6.331464138074493e-05, "epoch": 3.308714918759232, "percentage": 41.42, "elapsed_time": "2:38:17", "remaining_time": "3:43:52"}
+{"current_steps": 570, "total_steps": 1352, "loss": 0.2176, "learning_rate": 6.219126736369903e-05, "epoch": 3.3677991137370755, "percentage": 42.16, "elapsed_time": "2:41:06", "remaining_time": "3:41:01"}
+{"current_steps": 580, "total_steps": 1352, "loss": 0.2215, "learning_rate": 6.106131107689599e-05, "epoch": 3.4268833087149186, "percentage": 42.9, "elapsed_time": "2:43:52", "remaining_time": "3:38:06"}
+{"current_steps": 590, "total_steps": 1352, "loss": 0.2153, "learning_rate": 5.9925382602689974e-05, "epoch": 3.485967503692762, "percentage": 43.64, "elapsed_time": "2:46:38", "remaining_time": "3:35:13"}
+{"current_steps": 600, "total_steps": 1352, "loss": 0.2133, "learning_rate": 5.8784095247919305e-05, "epoch": 3.5450516986706058, "percentage": 44.38, "elapsed_time": "2:49:26", "remaining_time": "3:32:21"}
+{"current_steps": 610, "total_steps": 1352, "loss": 0.2109, "learning_rate": 5.763806521277184e-05, "epoch": 3.604135893648449, "percentage": 45.12, "elapsed_time": "2:52:16", "remaining_time": "3:29:32"}
+{"current_steps": 620, "total_steps": 1352, "loss": 0.2214, "learning_rate": 5.648791125808809e-05, "epoch": 3.6632200886262924, "percentage": 45.86, "elapsed_time": "2:55:02", "remaining_time": "3:26:39"}
+{"current_steps": 630, "total_steps": 1352, "loss": 0.212, "learning_rate": 5.5334254371281934e-05, "epoch": 3.722304283604136, "percentage": 46.6, "elapsed_time": "2:57:52", "remaining_time": "3:23:51"}
+{"current_steps": 640, "total_steps": 1352, "loss": 0.2196, "learning_rate": 5.417771743105907e-05, "epoch": 3.781388478581979, "percentage": 47.34, "elapsed_time": "3:00:45", "remaining_time": "3:21:05"}
+{"current_steps": 650, "total_steps": 1352, "loss": 0.2145, "learning_rate": 5.3018924871114305e-05, "epoch": 3.8404726735598227, "percentage": 48.08, "elapsed_time": "3:03:33", "remaining_time": "3:18:14"}
+{"current_steps": 660, "total_steps": 1352, "loss": 0.2199, "learning_rate": 5.185850234298942e-05, "epoch": 3.8995568685376663, "percentage": 48.82, "elapsed_time": "3:06:17", "remaining_time": "3:15:19"}
+{"current_steps": 670, "total_steps": 1352, "loss": 0.218, "learning_rate": 5.0697076378273354e-05, "epoch": 3.9586410635155094, "percentage": 49.56, "elapsed_time": "3:09:04", "remaining_time": "3:12:27"}
+{"current_steps": 680, "total_steps": 1352, "loss": 0.1987, "learning_rate": 4.953527405032723e-05, "epoch": 4.017725258493353, "percentage": 50.3, "elapsed_time": "3:11:51", "remaining_time": "3:09:35"}
+{"current_steps": 690, "total_steps": 1352, "loss": 0.1425, "learning_rate": 4.8373722635717086e-05, "epoch": 4.0768094534711965, "percentage": 51.04, "elapsed_time": "3:14:39", "remaining_time": "3:06:45"}
+{"current_steps": 700, "total_steps": 1352, "loss": 0.1313, "learning_rate": 4.721304927553658e-05, "epoch": 4.13589364844904, "percentage": 51.78, "elapsed_time": "3:17:26", "remaining_time": "3:03:53"}
+{"current_steps": 710, "total_steps": 1352, "loss": 0.1397, "learning_rate": 4.60538806368031e-05, "epoch": 4.194977843426884, "percentage": 52.51, "elapsed_time": "3:20:11", "remaining_time": "3:01:01"}
+{"current_steps": 720, "total_steps": 1352, "loss": 0.1421, "learning_rate": 4.489684257410958e-05, "epoch": 4.254062038404727, "percentage": 53.25, "elapsed_time": "3:22:58", "remaining_time": "2:58:09"}
+{"current_steps": 730, "total_steps": 1352, "loss": 0.1386, "learning_rate": 4.374255979171538e-05, "epoch": 4.31314623338257, "percentage": 53.99, "elapsed_time": "3:25:45", "remaining_time": "2:55:19"}
+{"current_steps": 740, "total_steps": 1352, "loss": 0.1444, "learning_rate": 4.2591655506257645e-05, "epoch": 4.372230428360414, "percentage": 54.73, "elapsed_time": "3:28:35", "remaining_time": "2:52:30"}
+{"current_steps": 750, "total_steps": 1352, "loss": 0.1391, "learning_rate": 4.144475111026643e-05, "epoch": 4.431314623338257, "percentage": 55.47, "elapsed_time": "3:31:25", "remaining_time": "2:49:42"}
+{"current_steps": 760, "total_steps": 1352, "loss": 0.1438, "learning_rate": 4.030246583666437e-05, "epoch": 4.4903988183161, "percentage": 56.21, "elapsed_time": "3:34:15", "remaining_time": "2:46:53"}
+{"current_steps": 770, "total_steps": 1352, "loss": 0.1415, "learning_rate": 3.9165416424432414e-05, "epoch": 4.549483013293944, "percentage": 56.95, "elapsed_time": "3:37:04", "remaining_time": "2:44:04"}
+{"current_steps": 780, "total_steps": 1352, "loss": 0.1492, "learning_rate": 3.803421678562213e-05, "epoch": 4.608567208271787, "percentage": 57.69, "elapsed_time": "3:39:52", "remaining_time": "2:41:14"}
+{"current_steps": 790, "total_steps": 1352, "loss": 0.1512, "learning_rate": 3.690947767389426e-05, "epoch": 4.66765140324963, "percentage": 58.43, "elapsed_time": "3:42:44", "remaining_time": "2:38:27"}
+{"current_steps": 800, "total_steps": 1352, "loss": 0.1481, "learning_rate": 3.57918063547627e-05, "epoch": 4.726735598227474, "percentage": 59.17, "elapsed_time": "3:45:31", "remaining_time": "2:35:36"}
+{"current_steps": 810, "total_steps": 1352, "loss": 0.1418, "learning_rate": 3.468180627772144e-05, "epoch": 4.7858197932053175, "percentage": 59.91, "elapsed_time": "3:48:15", "remaining_time": "2:32:44"}
+{"current_steps": 820, "total_steps": 1352, "loss": 0.146, "learning_rate": 3.358007675043224e-05, "epoch": 4.844903988183161, "percentage": 60.65, "elapsed_time": "3:51:03", "remaining_time": "2:29:54"}
+{"current_steps": 830, "total_steps": 1352, "loss": 0.1407, "learning_rate": 3.2487212615148316e-05, "epoch": 4.903988183161005, "percentage": 61.39, "elapsed_time": "3:53:49", "remaining_time": "2:27:03"}
+{"current_steps": 840, "total_steps": 1352, "loss": 0.1502, "learning_rate": 3.1403803927549006e-05, "epoch": 4.963072378138848, "percentage": 62.13, "elapsed_time": "3:56:38", "remaining_time": "2:24:14"}
+{"current_steps": 850, "total_steps": 1352, "loss": 0.1322, "learning_rate": 3.0330435638158806e-05, "epoch": 5.022156573116692, "percentage": 62.87, "elapsed_time": "3:59:24", "remaining_time": "2:21:23"}
+{"current_steps": 860, "total_steps": 1352, "loss": 0.0985, "learning_rate": 2.9267687276522876e-05, "epoch": 5.081240768094535, "percentage": 63.61, "elapsed_time": "4:02:13", "remaining_time": "2:18:34"}
+{"current_steps": 870, "total_steps": 1352, "loss": 0.0929, "learning_rate": 2.821613263830912e-05, "epoch": 5.140324963072378, "percentage": 64.35, "elapsed_time": "4:05:03", "remaining_time": "2:15:46"}
+{"current_steps": 880, "total_steps": 1352, "loss": 0.0941, "learning_rate": 2.717633947550651e-05, "epoch": 5.199409158050222, "percentage": 65.09, "elapsed_time": "4:07:51", "remaining_time": "2:12:56"}
+{"current_steps": 890, "total_steps": 1352, "loss": 0.0951, "learning_rate": 2.614886918988604e-05, "epoch": 5.258493353028065, "percentage": 65.83, "elapsed_time": "4:10:38", "remaining_time": "2:10:06"}
+{"current_steps": 900, "total_steps": 1352, "loss": 0.0926, "learning_rate": 2.5134276529890644e-05, "epoch": 5.317577548005908, "percentage": 66.57, "elapsed_time": "4:13:26", "remaining_time": "2:07:17"}
+{"current_steps": 910, "total_steps": 1352, "loss": 0.095, "learning_rate": 2.4133109291117156e-05, "epoch": 5.376661742983752, "percentage": 67.31, "elapsed_time": "4:16:14", "remaining_time": "2:04:27"}
+{"current_steps": 920, "total_steps": 1352, "loss": 0.0886, "learning_rate": 2.314590802055232e-05, "epoch": 5.435745937961595, "percentage": 68.05, "elapsed_time": "4:18:58", "remaining_time": "2:01:36"}
+{"current_steps": 930, "total_steps": 1352, "loss": 0.096, "learning_rate": 2.2173205724722318e-05, "epoch": 5.4948301329394384, "percentage": 68.79, "elapsed_time": "4:21:47", "remaining_time": "1:58:47"}
+{"current_steps": 940, "total_steps": 1352, "loss": 0.0962, "learning_rate": 2.121552758191366e-05, "epoch": 5.5539143279172825, "percentage": 69.53, "elapsed_time": "4:24:31", "remaining_time": "1:55:56"}
+{"current_steps": 950, "total_steps": 1352, "loss": 0.0985, "learning_rate": 2.027339065862064e-05, "epoch": 5.612998522895126, "percentage": 70.27, "elapsed_time": "4:27:16", "remaining_time": "1:53:05"}
+{"current_steps": 960, "total_steps": 1352, "loss": 0.0939, "learning_rate": 1.934730363037237e-05, "epoch": 5.672082717872969, "percentage": 71.01, "elapsed_time": "4:30:02", "remaining_time": "1:50:16"}
+{"current_steps": 970, "total_steps": 1352, "loss": 0.0975, "learning_rate": 1.843776650709046e-05, "epoch": 5.731166912850813, "percentage": 71.75, "elapsed_time": "4:32:47", "remaining_time": "1:47:25"}
+{"current_steps": 980, "total_steps": 1352, "loss": 0.093, "learning_rate": 1.7545270363125153e-05, "epoch": 5.790251107828656, "percentage": 72.49, "elapsed_time": "4:35:33", "remaining_time": "1:44:35"}
+{"current_steps": 990, "total_steps": 1352, "loss": 0.0959, "learning_rate": 1.6670297072116165e-05, "epoch": 5.849335302806499, "percentage": 73.22, "elapsed_time": "4:38:25", "remaining_time": "1:41:48"}
+{"current_steps": 1000, "total_steps": 1352, "loss": 0.0918, "learning_rate": 1.581331904682089e-05, "epoch": 5.908419497784343, "percentage": 73.96, "elapsed_time": "4:41:09", "remaining_time": "1:38:58"}
+{"current_steps": 1000, "total_steps": 1352, "eval_loss": 0.7311862707138062, "epoch": 5.908419497784343, "percentage": 73.96, "elapsed_time": "4:42:02", "remaining_time": "1:39:16"}
+{"current_steps": 1010, "total_steps": 1352, "loss": 0.0933, "learning_rate": 1.4974798984050942e-05, "epoch": 5.967503692762186, "percentage": 74.7, "elapsed_time": "4:45:04", "remaining_time": "1:36:31"}
+{"current_steps": 1020, "total_steps": 1352, "loss": 0.0828, "learning_rate": 1.4155189614854275e-05, "epoch": 6.026587887740029, "percentage": 75.44, "elapsed_time": "4:47:52", "remaining_time": "1:33:41"}
+{"current_steps": 1030, "total_steps": 1352, "loss": 0.0678, "learning_rate": 1.3354933460078217e-05, "epoch": 6.085672082717873, "percentage": 76.18, "elapsed_time": "4:50:41", "remaining_time": "1:30:52"}
+{"current_steps": 1040, "total_steps": 1352, "loss": 0.0629, "learning_rate": 1.257446259144494e-05, "epoch": 6.144756277695716, "percentage": 76.92, "elapsed_time": "4:53:27", "remaining_time": "1:28:02"}
+{"current_steps": 1050, "total_steps": 1352, "loss": 0.0697, "learning_rate": 1.1814198398268794e-05, "epoch": 6.203840472673559, "percentage": 77.66, "elapsed_time": "4:56:13", "remaining_time": "1:25:12"}
+{"current_steps": 1060, "total_steps": 1352, "loss": 0.0644, "learning_rate": 1.1074551359941021e-05, "epoch": 6.262924667651403, "percentage": 78.4, "elapsed_time": "4:59:01", "remaining_time": "1:22:22"}
+{"current_steps": 1070, "total_steps": 1352, "loss": 0.069, "learning_rate": 1.0355920824305127e-05, "epoch": 6.3220088626292466, "percentage": 79.14, "elapsed_time": "5:01:54", "remaining_time": "1:19:34"}
+{"current_steps": 1080, "total_steps": 1352, "loss": 0.0666, "learning_rate": 9.658694792042284e-06, "epoch": 6.381093057607091, "percentage": 79.88, "elapsed_time": "5:04:41", "remaining_time": "1:16:44"}
+{"current_steps": 1090, "total_steps": 1352, "loss": 0.0679, "learning_rate": 8.98324970718319e-06, "epoch": 6.440177252584934, "percentage": 80.62, "elapsed_time": "5:07:30", "remaining_time": "1:13:54"}
+{"current_steps": 1100, "total_steps": 1352, "loss": 0.0656, "learning_rate": 8.329950253859703e-06, "epoch": 6.499261447562777, "percentage": 81.36, "elapsed_time": "5:10:14", "remaining_time": "1:11:04"}
+{"current_steps": 1110, "total_steps": 1352, "loss": 0.0664, "learning_rate": 7.699149159405734e-06, "epoch": 6.558345642540621, "percentage": 82.1, "elapsed_time": "5:13:04", "remaining_time": "1:08:15"}
+{"current_steps": 1120, "total_steps": 1352, "loss": 0.0673, "learning_rate": 7.0911870039138015e-06, "epoch": 6.617429837518464, "percentage": 82.84, "elapsed_time": "5:15:53", "remaining_time": "1:05:26"}
+{"current_steps": 1130, "total_steps": 1352, "loss": 0.0697, "learning_rate": 6.506392036350167e-06, "epoch": 6.676514032496307, "percentage": 83.58, "elapsed_time": "5:18:38", "remaining_time": "1:02:36"}
+{"current_steps": 1140, "total_steps": 1352, "loss": 0.0669, "learning_rate": 5.945079997327713e-06, "epoch": 6.735598227474151, "percentage": 84.32, "elapsed_time": "5:21:24", "remaining_time": "0:59:46"}
+{"current_steps": 1150, "total_steps": 1352, "loss": 0.0683, "learning_rate": 5.407553948632277e-06, "epoch": 6.794682422451994, "percentage": 85.06, "elapsed_time": "5:24:09", "remaining_time": "0:56:56"}
+{"current_steps": 1160, "total_steps": 1352, "loss": 0.0684, "learning_rate": 4.894104109594466e-06, "epoch": 6.853766617429837, "percentage": 85.8, "elapsed_time": "5:26:58", "remaining_time": "0:54:07"}
+{"current_steps": 1170, "total_steps": 1352, "loss": 0.0687, "learning_rate": 4.405007700395497e-06, "epoch": 6.912850812407681, "percentage": 86.54, "elapsed_time": "5:29:44", "remaining_time": "0:51:17"}
+{"current_steps": 1180, "total_steps": 1352, "loss": 0.0721, "learning_rate": 3.940528792391223e-06, "epoch": 6.971935007385524, "percentage": 87.28, "elapsed_time": "5:32:25", "remaining_time": "0:48:27"}
+{"current_steps": 1190, "total_steps": 1352, "loss": 0.0653, "learning_rate": 3.5009181655356826e-06, "epoch": 7.0310192023633675, "percentage": 88.02, "elapsed_time": "5:35:13", "remaining_time": "0:45:38"}
+{"current_steps": 1200, "total_steps": 1352, "loss": 0.0579, "learning_rate": 3.0864131729807398e-06, "epoch": 7.0901033973412115, "percentage": 88.76, "elapsed_time": "5:38:03", "remaining_time": "0:42:49"}
+{"current_steps": 1210, "total_steps": 1352, "loss": 0.0585, "learning_rate": 2.6972376129251686e-06, "epoch": 7.149187592319055, "percentage": 89.5, "elapsed_time": "5:40:48", "remaining_time": "0:39:59"}
+{"current_steps": 1220, "total_steps": 1352, "loss": 0.0562, "learning_rate": 2.3336016077822154e-06, "epoch": 7.208271787296898, "percentage": 90.24, "elapsed_time": "5:43:37", "remaining_time": "0:37:10"}
+{"current_steps": 1230, "total_steps": 1352, "loss": 0.0572, "learning_rate": 1.9957014907310224e-06, "epoch": 7.267355982274742, "percentage": 90.98, "elapsed_time": "5:46:23", "remaining_time": "0:34:21"}
+{"current_steps": 1240, "total_steps": 1352, "loss": 0.06, "learning_rate": 1.6837196997130434e-06, "epoch": 7.326440177252585, "percentage": 91.72, "elapsed_time": "5:49:09", "remaining_time": "0:31:32"}
+{"current_steps": 1250, "total_steps": 1352, "loss": 0.0565, "learning_rate": 1.3978246789307149e-06, "epoch": 7.385524372230428, "percentage": 92.46, "elapsed_time": "5:51:56", "remaining_time": "0:28:43"}
+{"current_steps": 1260, "total_steps": 1352, "loss": 0.0562, "learning_rate": 1.1381707879016157e-06, "epoch": 7.444608567208272, "percentage": 93.2, "elapsed_time": "5:54:44", "remaining_time": "0:25:54"}
+{"current_steps": 1270, "total_steps": 1352, "loss": 0.0556, "learning_rate": 9.048982181171894e-07, "epoch": 7.503692762186115, "percentage": 93.93, "elapsed_time": "5:57:33", "remaining_time": "0:23:05"}
+{"current_steps": 1280, "total_steps": 1352, "loss": 0.0525, "learning_rate": 6.98132917350991e-07, "epoch": 7.562776957163958, "percentage": 94.67, "elapsed_time": "6:00:22", "remaining_time": "0:20:16"}
+{"current_steps": 1290, "total_steps": 1352, "loss": 0.0549, "learning_rate": 5.179865216573654e-07, "epoch": 7.621861152141802, "percentage": 95.41, "elapsed_time": "6:03:03", "remaining_time": "0:17:26"}
+{"current_steps": 1300, "total_steps": 1352, "loss": 0.0583, "learning_rate": 3.6455629509730136e-07, "epoch": 7.680945347119645, "percentage": 96.15, "elapsed_time": "6:05:55", "remaining_time": "0:14:38"}
+{"current_steps": 1310, "total_steps": 1352, "loss": 0.0553, "learning_rate": 2.3792507722388835e-07, "epoch": 7.7400295420974885, "percentage": 96.89, "elapsed_time": "6:08:39", "remaining_time": "0:11:49"}
+{"current_steps": 1320, "total_steps": 1352, "loss": 0.0556, "learning_rate": 1.3816123835588834e-07, "epoch": 7.7991137370753325, "percentage": 97.63, "elapsed_time": "6:11:26", "remaining_time": "0:09:00"}
+{"current_steps": 1330, "total_steps": 1352, "loss": 0.0589, "learning_rate": 6.531864266343113e-08, "epoch": 7.858197932053176, "percentage": 98.37, "elapsed_time": "6:14:11", "remaining_time": "0:06:11"}
+{"current_steps": 1340, "total_steps": 1352, "loss": 0.0603, "learning_rate": 1.943661908586636e-08, "epoch": 7.917282127031019, "percentage": 99.11, "elapsed_time": "6:17:00", "remaining_time": "0:03:22"}
+{"current_steps": 1350, "total_steps": 1352, "loss": 0.0568, "learning_rate": 5.399400973882251e-10, "epoch": 7.976366322008863, "percentage": 99.85, "elapsed_time": "6:19:47", "remaining_time": "0:00:33"}
+{"current_steps": 1352, "total_steps": 1352, "epoch": 7.988183161004431, "percentage": 100.0, "elapsed_time": "6:20:18", "remaining_time": "0:00:00"}

trainer_state.json

@@ -0,0 +1,1003 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.988183161004431,
+  "eval_steps": 500,
+  "global_step": 1352,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.059084194977843424,
+      "grad_norm": 0.6387189059092414,
+      "learning_rate": 9.998650208062712e-05,
+      "loss": 0.6972,
+      "step": 10
+    },
+    {
+      "epoch": 0.11816838995568685,
+      "grad_norm": 0.48601036335500336,
+      "learning_rate": 9.994601561026155e-05,
+      "loss": 0.5423,
+      "step": 20
+    },
+    {
+      "epoch": 0.17725258493353027,
+      "grad_norm": 0.40437591619558344,
+      "learning_rate": 9.98785624482278e-05,
+      "loss": 0.5149,
+      "step": 30
+    },
+    {
+      "epoch": 0.2363367799113737,
+      "grad_norm": 0.5030211090458666,
+      "learning_rate": 9.978417901361958e-05,
+      "loss": 0.513,
+      "step": 40
+    },
+    {
+      "epoch": 0.29542097488921715,
+      "grad_norm": 0.49098083558596206,
+      "learning_rate": 9.96629162656365e-05,
+      "loss": 0.4957,
+      "step": 50
+    },
+    {
+      "epoch": 0.35450516986706054,
+      "grad_norm": 0.6329012343929873,
+      "learning_rate": 9.951483967607041e-05,
+      "loss": 0.495,
+      "step": 60
+    },
+    {
+      "epoch": 0.413589364844904,
+      "grad_norm": 0.48252149699962754,
+      "learning_rate": 9.934002919395592e-05,
+      "loss": 0.4943,
+      "step": 70
+    },
+    {
+      "epoch": 0.4726735598227474,
+      "grad_norm": 0.5221971068930835,
+      "learning_rate": 9.91385792024048e-05,
+      "loss": 0.4738,
+      "step": 80
+    },
+    {
+      "epoch": 0.5317577548005908,
+      "grad_norm": 0.42549092560536134,
+      "learning_rate": 9.891059846764679e-05,
+      "loss": 0.4565,
+      "step": 90
+    },
+    {
+      "epoch": 0.5908419497784343,
+      "grad_norm": 0.4606171719652433,
+      "learning_rate": 9.865621008030492e-05,
+      "loss": 0.4674,
+      "step": 100
+    },
+    {
+      "epoch": 0.6499261447562777,
+      "grad_norm": 0.46353540699946943,
+      "learning_rate": 9.83755513889369e-05,
+      "loss": 0.4727,
+      "step": 110
+    },
+    {
+      "epoch": 0.7090103397341211,
+      "grad_norm": 0.4815928480268326,
+      "learning_rate": 9.80687739258782e-05,
+      "loss": 0.4736,
+      "step": 120
+    },
+    {
+      "epoch": 0.7680945347119645,
+      "grad_norm": 0.485838392040906,
+      "learning_rate": 9.773604332542729e-05,
+      "loss": 0.47,
+      "step": 130
+    },
+    {
+      "epoch": 0.827178729689808,
+      "grad_norm": 0.5025850953484241,
+      "learning_rate": 9.737753923441688e-05,
+      "loss": 0.467,
+      "step": 140
+    },
+    {
+      "epoch": 0.8862629246676514,
+      "grad_norm": 0.45997226542102815,
+      "learning_rate": 9.69934552152196e-05,
+      "loss": 0.4522,
+      "step": 150
+    },
+    {
+      "epoch": 0.9453471196454948,
+      "grad_norm": 0.4546309634161405,
+      "learning_rate": 9.658399864124037e-05,
+      "loss": 0.4613,
+      "step": 160
+    },
+    {
+      "epoch": 1.0044313146233383,
+      "grad_norm": 0.6577670848472944,
+      "learning_rate": 9.61493905849521e-05,
+      "loss": 0.4641,
+      "step": 170
+    },
+    {
+      "epoch": 1.0635155096011817,
+      "grad_norm": 0.5743711541659764,
+      "learning_rate": 9.568986569853487e-05,
+      "loss": 0.3946,
+      "step": 180
+    },
+    {
+      "epoch": 1.122599704579025,
+      "grad_norm": 0.5374571463070855,
+      "learning_rate": 9.520567208718337e-05,
+      "loss": 0.3882,
+      "step": 190
+    },
+    {
+      "epoch": 1.1816838995568686,
+      "grad_norm": 0.6531134738909915,
+      "learning_rate": 9.469707117515067e-05,
+      "loss": 0.4205,
+      "step": 200
+    },
+    {
+      "epoch": 1.240768094534712,
+      "grad_norm": 0.5201345622237219,
+      "learning_rate": 9.416433756460091e-05,
+      "loss": 0.386,
+      "step": 210
+    },
+    {
+      "epoch": 1.2998522895125553,
+      "grad_norm": 0.7267530776250423,
+      "learning_rate": 9.360775888734698e-05,
+      "loss": 0.4096,
+      "step": 220
+    },
+    {
+      "epoch": 1.3589364844903988,
+      "grad_norm": 0.5480147887972682,
+      "learning_rate": 9.302763564955331e-05,
+      "loss": 0.3921,
+      "step": 230
+    },
+    {
+      "epoch": 1.4180206794682422,
+      "grad_norm": 0.6228502280992079,
+      "learning_rate": 9.242428106948749e-05,
+      "loss": 0.3788,
+      "step": 240
+    },
+    {
+      "epoch": 1.4771048744460857,
+      "grad_norm": 0.7119116813192373,
+      "learning_rate": 9.179802090840853e-05,
+      "loss": 0.3894,
+      "step": 250
+    },
+    {
+      "epoch": 1.536189069423929,
+      "grad_norm": 0.5987495739217404,
+      "learning_rate": 9.114919329468282e-05,
+      "loss": 0.3707,
+      "step": 260
+    },
+    {
+      "epoch": 1.5952732644017726,
+      "grad_norm": 0.6278387167291306,
+      "learning_rate": 9.04781485412231e-05,
+      "loss": 0.3938,
+      "step": 270
+    },
+    {
+      "epoch": 1.654357459379616,
+      "grad_norm": 0.5312391005336979,
+      "learning_rate": 8.978524895634842e-05,
+      "loss": 0.3799,
+      "step": 280
+    },
+    {
+      "epoch": 1.7134416543574593,
+      "grad_norm": 0.7329625126762797,
+      "learning_rate": 8.907086864816803e-05,
+      "loss": 0.403,
+      "step": 290
+    },
+    {
+      "epoch": 1.7725258493353029,
+      "grad_norm": 0.5600401734580108,
+      "learning_rate": 8.833539332259398e-05,
+      "loss": 0.3758,
+      "step": 300
+    },
+    {
+      "epoch": 1.8316100443131462,
+      "grad_norm": 0.503395394356135,
+      "learning_rate": 8.757922007509207e-05,
+      "loss": 0.3963,
+      "step": 310
+    },
+    {
+      "epoch": 1.8906942392909896,
+      "grad_norm": 0.6324775953771359,
+      "learning_rate": 8.680275717628337e-05,
+      "loss": 0.3858,
+      "step": 320
+    },
+    {
+      "epoch": 1.9497784342688331,
+      "grad_norm": 0.5574107174736728,
+      "learning_rate": 8.600642385151205e-05,
+      "loss": 0.3799,
+      "step": 330
+    },
+    {
+      "epoch": 2.0088626292466767,
+      "grad_norm": 0.5250864362886176,
+      "learning_rate": 8.519065005449858e-05,
+      "loss": 0.3763,
+      "step": 340
+    },
+    {
+      "epoch": 2.06794682422452,
+      "grad_norm": 0.7754155552722259,
+      "learning_rate": 8.43558762352005e-05,
+      "loss": 0.2934,
+      "step": 350
+    },
+    {
+      "epoch": 2.1270310192023634,
+      "grad_norm": 0.6889925166837503,
+      "learning_rate": 8.350255310200612e-05,
+      "loss": 0.3078,
+      "step": 360
+    },
+    {
+      "epoch": 2.186115214180207,
+      "grad_norm": 0.6281519102970855,
+      "learning_rate": 8.263114137838947e-05,
+      "loss": 0.3028,
+      "step": 370
+    },
+    {
+      "epoch": 2.24519940915805,
+      "grad_norm": 0.7619069663070173,
+      "learning_rate": 8.174211155415799e-05,
+      "loss": 0.2972,
+      "step": 380
+    },
+    {
+      "epoch": 2.3042836041358936,
+      "grad_norm": 0.6469214076306162,
+      "learning_rate": 8.083594363142717e-05,
+      "loss": 0.2995,
+      "step": 390
+    },
+    {
+      "epoch": 2.363367799113737,
+      "grad_norm": 0.7761443706661119,
+      "learning_rate": 7.991312686545937e-05,
+      "loss": 0.2963,
+      "step": 400
+    },
+    {
+      "epoch": 2.4224519940915803,
+      "grad_norm": 0.6652219653359559,
+      "learning_rate": 7.897415950050676e-05,
+      "loss": 0.2987,
+      "step": 410
+    },
+    {
+      "epoch": 2.481536189069424,
+      "grad_norm": 0.7586700003856579,
+      "learning_rate": 7.801954850080075e-05,
+      "loss": 0.3092,
+      "step": 420
+    },
+    {
+      "epoch": 2.5406203840472674,
+      "grad_norm": 0.6663785963305503,
+      "learning_rate": 7.704980927683359e-05,
+      "loss": 0.2951,
+      "step": 430
+    },
+    {
+      "epoch": 2.5997045790251105,
+      "grad_norm": 0.6913254608481674,
+      "learning_rate": 7.60654654070796e-05,
+      "loss": 0.3097,
+      "step": 440
+    },
+    {
+      "epoch": 2.658788774002954,
+      "grad_norm": 0.6943652391419598,
+      "learning_rate": 7.506704835530634e-05,
+      "loss": 0.2999,
+      "step": 450
+    },
+    {
+      "epoch": 2.7178729689807977,
+      "grad_norm": 0.644096596125316,
+      "learning_rate": 7.405509718362842e-05,
+      "loss": 0.2905,
+      "step": 460
+    },
+    {
+      "epoch": 2.7769571639586412,
+      "grad_norm": 0.7660517390503399,
+      "learning_rate": 7.303015826145885e-05,
+      "loss": 0.309,
+      "step": 470
+    },
+    {
+      "epoch": 2.8360413589364843,
+      "grad_norm": 0.7708918433168639,
+      "learning_rate": 7.199278497051498e-05,
+      "loss": 0.302,
+      "step": 480
+    },
+    {
+      "epoch": 2.895125553914328,
+      "grad_norm": 0.626692120748867,
+      "learning_rate": 7.094353740603839e-05,
+      "loss": 0.297,
+      "step": 490
+    },
+    {
+      "epoch": 2.9542097488921715,
+      "grad_norm": 0.7926892798861292,
+      "learning_rate": 6.988298207439021e-05,
+      "loss": 0.3101,
+      "step": 500
+    },
+    {
+      "epoch": 2.9542097488921715,
+      "eval_loss": 0.4939613938331604,
+      "eval_runtime": 53.5552,
+      "eval_samples_per_second": 2.073,
+      "eval_steps_per_second": 0.523,
+      "step": 500
+    },
+    {
+      "epoch": 3.0132939438700146,
+      "grad_norm": 0.5747285632843896,
+      "learning_rate": 6.881169158718474e-05,
+      "loss": 0.2736,
+      "step": 510
+    },
+    {
+      "epoch": 3.072378138847858,
+      "grad_norm": 0.7661842329328333,
+      "learning_rate": 6.773024435212678e-05,
+      "loss": 0.2187,
+      "step": 520
+    },
+    {
+      "epoch": 3.1314623338257017,
+      "grad_norm": 0.8401289392682185,
+      "learning_rate": 6.663922426071977e-05,
+      "loss": 0.2057,
+      "step": 530
+    },
+    {
+      "epoch": 3.1905465288035453,
+      "grad_norm": 0.7506650372127406,
+      "learning_rate": 6.553922037301283e-05,
+      "loss": 0.2067,
+      "step": 540
+    },
+    {
+      "epoch": 3.2496307237813884,
+      "grad_norm": 0.7842941173009434,
+      "learning_rate": 6.443082659955738e-05,
+      "loss": 0.1989,
+      "step": 550
+    },
+    {
+      "epoch": 3.308714918759232,
+      "grad_norm": 0.7593744868915973,
+      "learning_rate": 6.331464138074493e-05,
+      "loss": 0.2179,
+      "step": 560
+    },
+    {
+      "epoch": 3.3677991137370755,
+      "grad_norm": 0.756091123285611,
+      "learning_rate": 6.219126736369903e-05,
+      "loss": 0.2176,
+      "step": 570
+    },
+    {
+      "epoch": 3.4268833087149186,
+      "grad_norm": 0.7869512689224245,
+      "learning_rate": 6.106131107689599e-05,
+      "loss": 0.2215,
+      "step": 580
+    },
+    {
+      "epoch": 3.485967503692762,
+      "grad_norm": 0.7636724676762149,
+      "learning_rate": 5.9925382602689974e-05,
+      "loss": 0.2153,
+      "step": 590
+    },
+    {
+      "epoch": 3.5450516986706058,
+      "grad_norm": 0.6815318566179079,
+      "learning_rate": 5.8784095247919305e-05,
+      "loss": 0.2133,
+      "step": 600
+    },
+    {
+      "epoch": 3.604135893648449,
+      "grad_norm": 0.7848948829508617,
+      "learning_rate": 5.763806521277184e-05,
+      "loss": 0.2109,
+      "step": 610
+    },
+    {
+      "epoch": 3.6632200886262924,
+      "grad_norm": 0.7715200213470335,
+      "learning_rate": 5.648791125808809e-05,
+      "loss": 0.2214,
+      "step": 620
+    },
+    {
+      "epoch": 3.722304283604136,
+      "grad_norm": 0.6480973448749833,
+      "learning_rate": 5.5334254371281934e-05,
+      "loss": 0.212,
+      "step": 630
+    },
+    {
+      "epoch": 3.781388478581979,
+      "grad_norm": 0.6618796234383102,
+      "learning_rate": 5.417771743105907e-05,
+      "loss": 0.2196,
+      "step": 640
+    },
+    {
+      "epoch": 3.8404726735598227,
+      "grad_norm": 0.9462710180746308,
+      "learning_rate": 5.3018924871114305e-05,
+      "loss": 0.2145,
+      "step": 650
+    },
+    {
+      "epoch": 3.8995568685376663,
+      "grad_norm": 0.7674586984793125,
+      "learning_rate": 5.185850234298942e-05,
+      "loss": 0.2199,
+      "step": 660
+    },
+    {
+      "epoch": 3.9586410635155094,
+      "grad_norm": 0.8420752213128357,
+      "learning_rate": 5.0697076378273354e-05,
+      "loss": 0.218,
+      "step": 670
+    },
+    {
+      "epoch": 4.017725258493353,
+      "grad_norm": 0.6235542870137168,
+      "learning_rate": 4.953527405032723e-05,
+      "loss": 0.1987,
+      "step": 680
+    },
+    {
+      "epoch": 4.0768094534711965,
+      "grad_norm": 0.8247620064480204,
+      "learning_rate": 4.8373722635717086e-05,
+      "loss": 0.1425,
+      "step": 690
+    },
+    {
+      "epoch": 4.13589364844904,
+      "grad_norm": 0.806070554149328,
+      "learning_rate": 4.721304927553658e-05,
+      "loss": 0.1313,
+      "step": 700
+    },
+    {
+      "epoch": 4.194977843426884,
+      "grad_norm": 1.1960115947293068,
+      "learning_rate": 4.60538806368031e-05,
+      "loss": 0.1397,
+      "step": 710
+    },
+    {
+      "epoch": 4.254062038404727,
+      "grad_norm": 0.7571134276917469,
+      "learning_rate": 4.489684257410958e-05,
+      "loss": 0.1421,
+      "step": 720
+    },
+    {
+      "epoch": 4.31314623338257,
+      "grad_norm": 0.8865025891479803,
+      "learning_rate": 4.374255979171538e-05,
+      "loss": 0.1386,
+      "step": 730
+    },
+    {
+      "epoch": 4.372230428360414,
+      "grad_norm": 0.9290220213968287,
+      "learning_rate": 4.2591655506257645e-05,
+      "loss": 0.1444,
+      "step": 740
+    },
+    {
+      "epoch": 4.431314623338257,
+      "grad_norm": 0.8896418597603776,
+      "learning_rate": 4.144475111026643e-05,
+      "loss": 0.1391,
+      "step": 750
+    },
+    {
+      "epoch": 4.4903988183161,
+      "grad_norm": 0.9053211649213782,
+      "learning_rate": 4.030246583666437e-05,
+      "loss": 0.1438,
+      "step": 760
+    },
+    {
+      "epoch": 4.549483013293944,
+      "grad_norm": 0.8453862385052026,
+      "learning_rate": 3.9165416424432414e-05,
+      "loss": 0.1415,
+      "step": 770
+    },
+    {
+      "epoch": 4.608567208271787,
+      "grad_norm": 0.8724405655899441,
+      "learning_rate": 3.803421678562213e-05,
+      "loss": 0.1492,
+      "step": 780
+    },
+    {
+      "epoch": 4.66765140324963,
+      "grad_norm": 0.98302153579186,
+      "learning_rate": 3.690947767389426e-05,
+      "loss": 0.1512,
+      "step": 790
+    },
+    {
+      "epoch": 4.726735598227474,
+      "grad_norm": 0.8137008256198869,
+      "learning_rate": 3.57918063547627e-05,
+      "loss": 0.1481,
+      "step": 800
+    },
+    {
+      "epoch": 4.7858197932053175,
+      "grad_norm": 0.8604546990181478,
+      "learning_rate": 3.468180627772144e-05,
+      "loss": 0.1418,
+      "step": 810
+    },
+    {
+      "epoch": 4.844903988183161,
+      "grad_norm": 0.8028241587093687,
+      "learning_rate": 3.358007675043224e-05,
+      "loss": 0.146,
+      "step": 820
+    },
+    {
+      "epoch": 4.903988183161005,
+      "grad_norm": 0.8209334713352179,
+      "learning_rate": 3.2487212615148316e-05,
+      "loss": 0.1407,
+      "step": 830
+    },
+    {
+      "epoch": 4.963072378138848,
+      "grad_norm": 0.9318035909918932,
+      "learning_rate": 3.1403803927549006e-05,
+      "loss": 0.1502,
+      "step": 840
+    },
+    {
+      "epoch": 5.022156573116692,
+      "grad_norm": 0.6982555180872486,
+      "learning_rate": 3.0330435638158806e-05,
+      "loss": 0.1322,
+      "step": 850
+    },
+    {
+      "epoch": 5.081240768094535,
+      "grad_norm": 0.9099529668785485,
+      "learning_rate": 2.9267687276522876e-05,
+      "loss": 0.0985,
+      "step": 860
+    },
+    {
+      "epoch": 5.140324963072378,
+      "grad_norm": 0.8093436219539469,
+      "learning_rate": 2.821613263830912e-05,
+      "loss": 0.0929,
+      "step": 870
+    },
+    {
+      "epoch": 5.199409158050222,
+      "grad_norm": 0.908675419764728,
+      "learning_rate": 2.717633947550651e-05,
+      "loss": 0.0941,
+      "step": 880
+    },
+    {
+      "epoch": 5.258493353028065,
+      "grad_norm": 0.9144782506889362,
+      "learning_rate": 2.614886918988604e-05,
+      "loss": 0.0951,
+      "step": 890
+    },
+    {
+      "epoch": 5.317577548005908,
+      "grad_norm": 0.8302439773379418,
+      "learning_rate": 2.5134276529890644e-05,
+      "loss": 0.0926,
+      "step": 900
+    },
+    {
+      "epoch": 5.376661742983752,
+      "grad_norm": 0.8145999759475107,
+      "learning_rate": 2.4133109291117156e-05,
+      "loss": 0.095,
+      "step": 910
+    },
+    {
+      "epoch": 5.435745937961595,
+      "grad_norm": 0.7459552675334057,
+      "learning_rate": 2.314590802055232e-05,
+      "loss": 0.0886,
+      "step": 920
+    },
+    {
+      "epoch": 5.4948301329394384,
+      "grad_norm": 0.9033419203896008,
+      "learning_rate": 2.2173205724722318e-05,
+      "loss": 0.096,
+      "step": 930
+    },
+    {
+      "epoch": 5.5539143279172825,
+      "grad_norm": 0.882377183289936,
+      "learning_rate": 2.121552758191366e-05,
+      "loss": 0.0962,
+      "step": 940
+    },
+    {
+      "epoch": 5.612998522895126,
+      "grad_norm": 0.8033415953079253,
+      "learning_rate": 2.027339065862064e-05,
+      "loss": 0.0985,
+      "step": 950
+    },
+    {
+      "epoch": 5.672082717872969,
+      "grad_norm": 0.9269892443045443,
+      "learning_rate": 1.934730363037237e-05,
+      "loss": 0.0939,
+      "step": 960
+    },
+    {
+      "epoch": 5.731166912850813,
+      "grad_norm": 0.8483635995759108,
+      "learning_rate": 1.843776650709046e-05,
+      "loss": 0.0975,
+      "step": 970
+    },
+    {
+      "epoch": 5.790251107828656,
+      "grad_norm": 0.7664349269568396,
+      "learning_rate": 1.7545270363125153e-05,
+      "loss": 0.093,
+      "step": 980
+    },
+    {
+      "epoch": 5.849335302806499,
+      "grad_norm": 0.7270945706246518,
+      "learning_rate": 1.6670297072116165e-05,
+      "loss": 0.0959,
+      "step": 990
+    },
+    {
+      "epoch": 5.908419497784343,
+      "grad_norm": 0.4763779090409354,
+      "learning_rate": 1.581331904682089e-05,
+      "loss": 0.0918,
+      "step": 1000
+    },
+    {
+      "epoch": 5.908419497784343,
+      "eval_loss": 0.7311862707138062,
+      "eval_runtime": 52.641,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 0.532,
+      "step": 1000
+    },
+    {
+      "epoch": 5.967503692762186,
+      "grad_norm": 0.7880059330281542,
+      "learning_rate": 1.4974798984050942e-05,
+      "loss": 0.0933,
+      "step": 1010
+    },
+    {
+      "epoch": 6.026587887740029,
+      "grad_norm": 0.5434228078562925,
+      "learning_rate": 1.4155189614854275e-05,
+      "loss": 0.0828,
+      "step": 1020
+    },
+    {
+      "epoch": 6.085672082717873,
+      "grad_norm": 0.6734629959345333,
+      "learning_rate": 1.3354933460078217e-05,
+      "loss": 0.0678,
+      "step": 1030
+    },
+    {
+      "epoch": 6.144756277695716,
+      "grad_norm": 0.6754051572707525,
+      "learning_rate": 1.257446259144494e-05,
+      "loss": 0.0629,
+      "step": 1040
+    },
+    {
+      "epoch": 6.203840472673559,
+      "grad_norm": 0.6799650760230072,
+      "learning_rate": 1.1814198398268794e-05,
+      "loss": 0.0697,
+      "step": 1050
+    },
+    {
+      "epoch": 6.262924667651403,
+      "grad_norm": 0.6540795313123376,
+      "learning_rate": 1.1074551359941021e-05,
+      "loss": 0.0644,
+      "step": 1060
+    },
+    {
+      "epoch": 6.3220088626292466,
+      "grad_norm": 0.814973115457644,
+      "learning_rate": 1.0355920824305127e-05,
+      "loss": 0.069,
+      "step": 1070
+    },
+    {
+      "epoch": 6.381093057607091,
+      "grad_norm": 0.8090154097885933,
+      "learning_rate": 9.658694792042284e-06,
+      "loss": 0.0666,
+      "step": 1080
+    },
+    {
+      "epoch": 6.440177252584934,
+      "grad_norm": 0.7016078216972328,
+      "learning_rate": 8.98324970718319e-06,
+      "loss": 0.0679,
+      "step": 1090
+    },
+    {
+      "epoch": 6.499261447562777,
+      "grad_norm": 0.7288879724763213,
+      "learning_rate": 8.329950253859703e-06,
+      "loss": 0.0656,
+      "step": 1100
+    },
+    {
+      "epoch": 6.558345642540621,
+      "grad_norm": 0.7247263147770627,
+      "learning_rate": 7.699149159405734e-06,
+      "loss": 0.0664,
+      "step": 1110
+    },
+    {
+      "epoch": 6.617429837518464,
+      "grad_norm": 0.6219352284185693,
+      "learning_rate": 7.0911870039138015e-06,
+      "loss": 0.0673,
+      "step": 1120
+    },
+    {
+      "epoch": 6.676514032496307,
+      "grad_norm": 0.7481718513639776,
+      "learning_rate": 6.506392036350167e-06,
+      "loss": 0.0697,
+      "step": 1130
+    },
+    {
+      "epoch": 6.735598227474151,
+      "grad_norm": 0.7642809598885449,
+      "learning_rate": 5.945079997327713e-06,
+      "loss": 0.0669,
+      "step": 1140
+    },
+    {
+      "epoch": 6.794682422451994,
+      "grad_norm": 0.7060296135322504,
+      "learning_rate": 5.407553948632277e-06,
+      "loss": 0.0683,
+      "step": 1150
+    },
+    {
+      "epoch": 6.853766617429837,
+      "grad_norm": 0.7218214547563072,
+      "learning_rate": 4.894104109594466e-06,
+      "loss": 0.0684,
+      "step": 1160
+    },
+    {
+      "epoch": 6.912850812407681,
+      "grad_norm": 0.6944520001481999,
+      "learning_rate": 4.405007700395497e-06,
+      "loss": 0.0687,
+      "step": 1170
+    },
+    {
+      "epoch": 6.971935007385524,
+      "grad_norm": 0.8260640922228497,
+      "learning_rate": 3.940528792391223e-06,
+      "loss": 0.0721,
+      "step": 1180
+    },
+    {
+      "epoch": 7.0310192023633675,
+      "grad_norm": 0.6707004455651014,
+      "learning_rate": 3.5009181655356826e-06,
+      "loss": 0.0653,
+      "step": 1190
+    },
+    {
+      "epoch": 7.0901033973412115,
+      "grad_norm": 0.6805257469848828,
+      "learning_rate": 3.0864131729807398e-06,
+      "loss": 0.0579,
+      "step": 1200
+    },
+    {
+      "epoch": 7.149187592319055,
+      "grad_norm": 0.5495171159123228,
+      "learning_rate": 2.6972376129251686e-06,
+      "loss": 0.0585,
+      "step": 1210
+    },
+    {
+      "epoch": 7.208271787296898,
+      "grad_norm": 0.6136512130241976,
+      "learning_rate": 2.3336016077822154e-06,
+      "loss": 0.0562,
+      "step": 1220
+    },
+    {
+      "epoch": 7.267355982274742,
+      "grad_norm": 0.7591755724568,
+      "learning_rate": 1.9957014907310224e-06,
+      "loss": 0.0572,
+      "step": 1230
+    },
+    {
+      "epoch": 7.326440177252585,
+      "grad_norm": 0.6022538817881757,
+      "learning_rate": 1.6837196997130434e-06,
+      "loss": 0.06,
+      "step": 1240
+    },
+    {
+      "epoch": 7.385524372230428,
+      "grad_norm": 0.774444698651241,
+      "learning_rate": 1.3978246789307149e-06,
+      "loss": 0.0565,
+      "step": 1250
+    },
+    {
+      "epoch": 7.444608567208272,
+      "grad_norm": 0.7442025153147653,
+      "learning_rate": 1.1381707879016157e-06,
+      "loss": 0.0562,
+      "step": 1260
+    },
+    {
+      "epoch": 7.503692762186115,
+      "grad_norm": 0.5700289390004409,
+      "learning_rate": 9.048982181171894e-07,
+      "loss": 0.0556,
+      "step": 1270
+    },
+    {
+      "epoch": 7.562776957163958,
+      "grad_norm": 0.695724523693832,
+      "learning_rate": 6.98132917350991e-07,
+      "loss": 0.0525,
+      "step": 1280
+    },
+    {
+      "epoch": 7.621861152141802,
+      "grad_norm": 0.6867233808424225,
+      "learning_rate": 5.179865216573654e-07,
+      "loss": 0.0549,
+      "step": 1290
+    },
+    {
+      "epoch": 7.680945347119645,
+      "grad_norm": 0.7051791342497011,
+      "learning_rate": 3.6455629509730136e-07,
+      "loss": 0.0583,
+      "step": 1300
+    },
+    {
+      "epoch": 7.7400295420974885,
+      "grad_norm": 0.6772707185024253,
+      "learning_rate": 2.3792507722388835e-07,
+      "loss": 0.0553,
+      "step": 1310
+    },
+    {
+      "epoch": 7.7991137370753325,
+      "grad_norm": 0.6815226755163512,
+      "learning_rate": 1.3816123835588834e-07,
+      "loss": 0.0556,
+      "step": 1320
+    },
+    {
+      "epoch": 7.858197932053176,
+      "grad_norm": 0.6475756485632287,
+      "learning_rate": 6.531864266343113e-08,
+      "loss": 0.0589,
+      "step": 1330
+    },
+    {
+      "epoch": 7.917282127031019,
+      "grad_norm": 0.7007100620520975,
+      "learning_rate": 1.943661908586636e-08,
+      "loss": 0.0603,
+      "step": 1340
+    },
+    {
+      "epoch": 7.976366322008863,
+      "grad_norm": 0.7030879428421536,
+      "learning_rate": 5.399400973882251e-10,
+      "loss": 0.0568,
+      "step": 1350
+    },
+    {
+      "epoch": 7.988183161004431,
+      "step": 1352,
+      "total_flos": 1435586865135616.0,
+      "train_loss": 0.22072081432606167,
+      "train_runtime": 22818.9262,
+      "train_samples_per_second": 1.896,
+      "train_steps_per_second": 0.059
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1352,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 8,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1435586865135616.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:899f0486eced747a2f123ddafb9d6fecbc5fefa74f8054cc454aa966f7165bc2
+size 6968