See axolotl config

axolotl version: 0.10.0.dev0

adapter: lora
base_model: sethuiyer/Medichat-Llama3-8B
bf16: true
chat_template: llama3
datasets:
- data_files:
  - a7bec42aee25e0b5_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/
  type:
    field_input: input
    field_instruction: instruct
    field_output: output
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
eval_max_new_tokens: 256
evals_per_epoch: 2
flash_attention: false
fp16: false
gradient_accumulation_steps: 1
gradient_checkpointing: true
group_by_length: true
hub_model_id: apriasmoro/044d2362-b849-4351-ac66-7e5701ba5afd
learning_rate: 0.0002
logging_steps: 10
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: false
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_steps: 285
micro_batch_size: 4
mlflow_experiment_name: /tmp/a7bec42aee25e0b5_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
sample_packing: false
save_steps: 31
sequence_len: 2048
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: fd3933fe-ac58-454e-ae65-498874e457dd
wandb_project: Gradients-On-Demand
wandb_run: apriasmoro
wandb_runid: fd3933fe-ac58-454e-ae65-498874e457dd
warmup_steps: 100
weight_decay: 0.01

044d2362-b849-4351-ac66-7e5701ba5afd

This model is a fine-tuned version of sethuiyer/Medichat-Llama3-8B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3859

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: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• total_train_batch_size: 32
• total_eval_batch_size: 32
• 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
• training_steps: 285

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0023	1	0.6329
0.482	0.1088	48	0.4503
0.4092	0.2177	96	0.4176
0.4169	0.3265	144	0.4043
0.391	0.4354	192	0.3921
0.3942	0.5442	240	0.3859

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1