---
base_model: BanglaLLM/BanglaLLama-3.2-1b-unolp-culturax-base-v0.0.1
library_name: peft
tags:
- axolotl
- generated_from_trainer
model-index:
- name: BanglaLLama-3.2-1b-bangla-alpaca-orca-instruct-v0.0.1
  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. -->

[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.1`
```yaml
base_model: BanglaLLM/BanglaLLama-3.2-1b-unolp-culturax-base-v0.0.1
model_type: LlamaForCausalLM
  #adapter: /workspace/outputs/llama-3-8b-pretrain-uonlp-culturax-24Jul2024_1_04am/checkpoint-10000/adapter_model.safetensors
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: BanglaLLM/bangla-alpaca-orca 
    type: alpaca
  #max_steps: 10000
  #pretraining_dataset:
  #- path: "uonlp/CulturaX"
  #  name: bn
  #  type: pretrain
  #dataset_prepared_path: /workspace/datasets/last_run_prepared_uonlp_CulturaX_bn_24Jul2024_1_04am
val_set_size: 0.05
output_dir: /workspace/outputs/llama-3.2-1b-finetune-bangla-alpaca-orca-5Oct2024_10_00_AM

# push to hf hub
hub_model_id: BanglaLLM/BanglaLLama-3.2-1b-bangla-alpaca-orca-instruct-v0.0.1
hub_strategy: end

sequence_len: 4096
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
remove_unused_columns: true

  #peft: true
adapter: lora
lora_model_dir:
#lora_model_dir: /workspace/outputs/llama-3-8b-pretrain-uonlp-culturax-24Jul2024_1_04am/checkpoint-10000
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_modules_to_save:
  - embed_tokens
  - lm_head

wandb_project: banglallm-training
wandb_entity: banglallm
wandb_watch:
wandb_name: balglallm-training-llama3.2-1b-finetuning-5Oct2024_10_00_AM  
wandb_run_id: balglallm-training-llama3.2-1b-finetuning-5Oct2024_10_00_AM-id-1
wandb_log_model: checkpoint

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 2
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
   pad_token: <|end_of_text|>

```

</details><br>

[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/banglallm/banglallm-training/runs/balglallm-training-llama3.2-1b-finetuning-5Oct2024_10_00_AM-id-1)
# BanglaLLama-3.2-1b-bangla-alpaca-orca-instruct-v0.0.1

This model is a fine-tuned version of [BanglaLLM/BanglaLLama-3.2-1b-unolp-culturax-base-v0.0.1](https://huggingface.co/BanglaLLM/BanglaLLama-3.2-1b-unolp-culturax-base-v0.0.1) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4511

## 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: 0.0002
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 1

### Training results

| Training Loss | Epoch  | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| 0.8315        | 0.0002 | 1    | 0.9496          |
| 0.4253        | 0.2501 | 1424 | 0.5142          |
| 0.487         | 0.5003 | 2848 | 0.4754          |
| 0.4383        | 0.7504 | 4272 | 0.4550          |
| 0.478         | 1.0006 | 5696 | 0.4511          |


### Framework versions

- PEFT 0.12.0
- Transformers 4.44.2
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1