---
library_name: peft
license: apache-2.0
base_model: ibm-granite/granite-3.1-8b-instruct
tags:
- generated_from_trainer
model-index:
- name: home/ec2-user/SageMaker/task_decomposition/trained_models/granite-math-plans-3.1-8b-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. -->

[<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: `0.5.2`
```yaml
base_model: ibm-granite/granite-3.1-8b-instruct
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

resize_token_embeddings_to_32x: true
load_in_8bit: true
load_in_4bit: false
strict: false

datasets:
- path: /home/ec2-user/SageMaker/task_decomposition/data/task_decomposition_training_data_math.jsonl
  type: chat_template
  chat_template: tokenizer_default
  field_messages: conversations
  message_field_role: role
  message_field_content: value
dataset_prepared_path: last_run_prepared_sft

val_set_size: 0
sequence_len: 8192
sample_packing: false
pad_to_sequence_len: true
eval_sample_packing: false
output_dir: /home/ec2-user/SageMaker/task_decomposition/trained_models/granite-math-plans-3.1-8b-lora

wandb_project: null
wandb_entity: null
wandb_watch: null
wandb_name: null
wandb_log_model: null

adapter: lora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

gradient_accumulation_steps: 8
micro_batch_size: 1
eval_batch_size: 1
num_epochs: 3
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 1e-05

max_grad_norm: 1.0
logging_steps: 10

train_on_inputs: false
group_by_length: false

bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
xformers_attention:
flash_attention: true
warmup_ratio: 0.05
eval_steps:
save_strategy: epoch
eval_table_size:
num_processes: 8
deepspeed:
weight_decay: 0.0
```

</details><br>

# home/ec2-user/SageMaker/task_decomposition/trained_models/granite-math-plans-3.0-8b-lora

This model is a fine-tuned version of [ibm-granite/granite-3.1-8b-instruct](https://huggingface.co/ibm-granite/granite-3.1-8b-instruct) on the /home/ec2-user/SageMaker/task_decomposition/data/task_decomposition_training_data_math.jsonl dataset.

## 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: 1e-05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 64
- total_eval_batch_size: 8
- optimizer: Use adamw_bnb_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 154
- num_epochs: 3

### Training results



### Framework versions

- PEFT 0.13.2
- Transformers 4.46.3
- Pytorch 2.3.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3