llam3_esi

<!-- 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"/>

<details><summary>See axolotl config</summary>

axolotl version: 0.8.0.dev0

# Adapter & Model
adapter: lora
base_model: meta-llama/Meta-Llama-3-8B-Instruct
bf16: auto
load_in_8bit: true
special_tokens:
  pad_token: "<PAD>"

# Dataset
dataset_processes: 32
datasets:
  - path: /workspace/data/alpaca_esi_dataset.jsonl
    type: alpaca
    trust_remote_code: false
    message_property_mappings:
      instruction: instruction
      input: input
      output: output

# Output
output_dir: /workspace/data/outputs/llama3_esi

# Training Parameters
sequence_len: 1024
micro_batch_size: 64
gradient_accumulation_steps: 1
gradient_checkpointing: true
num_epochs: 3
learning_rate: 0.0002
optimizer: adamw_bnb_8bit

# LoRA Settings
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - gate_proj
  - down_proj
  - up_proj

# Trainer Settings
train_on_inputs: false
save_strategy: epoch
save_total_limit: 1
save_safetensors: true
logging_steps: 10
tokenizer_pad_to_eos_token: true

# Misc
shuffle_merged_datasets: true
skip_prepare_dataset: false
strict: false
ray_num_workers: 1
resources_per_worker:
  GPU: 1
use_ray: false
val_set_size: 0.0
weight_decay: 0.0

# TRL settings for compatibility
trl:
  log_completions: false
  ref_model_mixup_alpha: 0.9
  ref_model_sync_steps: 64
  sync_ref_model: false
  use_vllm: false
  vllm_device: auto
  vllm_dtype: auto
  vllm_gpu_memory_utilization: 0.9

</details><br>

workspace/data/outputs/llama3_esi

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the /workspace/data/alpaca_esi_dataset.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: 0.0002
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_BNB 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: 100
• num_epochs: 3.0

Training results

Framework versions

• PEFT 0.14.0
• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0