See axolotl config

axolotl version: 0.5.2

adapter: lora
base_model: unsloth/mistral-7b-instruct-v0.2
bf16: auto
chat_template: llama3
datasets:
- data_files:
  - f643fd0fe7e0bfb5_train_data.json
  ds_type: json
  format: custom
  path: /runs/taopanda-1_cae7d918-8995-47e5-995d-5bd97d626e49/f643fd0fe7e0bfb5_train_data.json
  preprocessing:
  - shuffle: true
  type:
    field: null
    field_input: chosen_gpt
    field_instruction: prompt_id
    field_output: rejected_gpt
    field_system: null
    format: null
    no_input_format: null
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device_map: auto
early_stopping_patience: null
eval_max_new_tokens: 128
eval_strategy: 'no'
fp16: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: true
hub_model_id: taopanda-1/ede5d1a0-546f-43e9-8879-f16e745c50f0
learning_rate: 0.000195548260923036
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.02
lora_fan_in_fan_out: true
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 1000
micro_batch_size: 8
model_type: AutoModelForCausalLM
num_epochs: 1
optimizer: paged_adamw_8bit
output_dir: ./outputs/lora-out/taopanda-1_cae7d918-8995-47e5-995d-5bd97d626e49
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
save_steps: 0.1
save_total_limit: 1
seed: 32892
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
wandb_entity: fatcat87-taopanda
wandb_mode: online
wandb_name: taopanda-1_cae7d918-8995-47e5-995d-5bd97d626e49
wandb_project: subnet56
wandb_runid: taopanda-1_cae7d918-8995-47e5-995d-5bd97d626e49
warmup_ratio: 0.1
weight_decay: 0.05
xformers_attention: null

ede5d1a0-546f-43e9-8879-f16e745c50f0

This model is a fine-tuned version of unsloth/mistral-7b-instruct-v0.2 on the None 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: 0.000195548260923036
• train_batch_size: 8
• eval_batch_size: 8
• seed: 32892
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• total_eval_batch_size: 32
• optimizer: Use OptimizerNames.PAGED_ADAMW_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: 48
• training_steps: 482

Training results

Framework versions

• PEFT 0.13.2
• Transformers 4.46.3
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3