See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Henrychur/MMed-Llama-3-8B-EnIns
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 04aca996ecd32f7d_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/04aca996ecd32f7d_train_data.json
  type:
    field_input: context
    field_instruction: question
    field_output: final_decision
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 4
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: false
hub_model_id: sn5601/2991cd9c-b266-40c8-811b-30dfd6e89b83
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0001
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lr_scheduler: cosine
max_steps: 30
micro_batch_size: 2
mlflow_experiment_name: /tmp/04aca996ecd32f7d_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
saves_per_epoch: 4
sequence_len: 1024
special_tokens:
  pad_token: <|end_of_text|>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: sn56-miner
wandb_mode: disabled
wandb_name: 2991cd9c-b266-40c8-811b-30dfd6e89b83
wandb_project: god
wandb_run: gt3s
wandb_runid: 2991cd9c-b266-40c8-811b-30dfd6e89b83
warmup_steps: 10
weight_decay: 0.01
xformers_attention: true

2991cd9c-b266-40c8-811b-30dfd6e89b83

This model is a fine-tuned version of Henrychur/MMed-Llama-3-8B-EnIns on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1088

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: 0.0001
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 10
• training_steps: 30

Training results

Training Loss	Epoch	Step	Validation Loss
12.0326	0.0000	1	12.3249
12.4453	0.0001	3	11.6343
6.8853	0.0002	6	4.2496
0.0052	0.0004	9	0.8712
1.9453	0.0005	12	0.1916
0.3239	0.0006	15	0.1434
0.2602	0.0007	18	0.1199
0.6821	0.0008	21	0.1454
0.0071	0.0010	24	0.0948
0.0037	0.0011	27	0.1050
0.0039	0.0012	30	0.1088

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1