See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: furiosa-ai/mlperf-gpt-j-6b
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 1c92e33d67cb4908_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/1c92e33d67cb4908_train_data.json
  type:
    field_instruction: input persona
    field_output: synthesized text
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 400
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/77246c17-9c78-4020-9b21-ace66756a40d
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
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
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 5328
micro_batch_size: 2
mlflow_experiment_name: /tmp/1c92e33d67cb4908_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 400
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.04
wandb_entity: null
wandb_mode: online
wandb_name: e51a9104-f9a0-4430-8a0c-2df2aa197b76
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: e51a9104-f9a0-4430-8a0c-2df2aa197b76
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

77246c17-9c78-4020-9b21-ace66756a40d

This model is a fine-tuned version of furiosa-ai/mlperf-gpt-j-6b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9732

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: 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: 5328

Training results

Training Loss	Epoch	Step	Validation Loss
8.171	0.0002	1	2.2453
5.0901	0.0683	400	1.1782
4.2801	0.1366	800	1.1248
4.4934	0.2049	1200	1.0934
4.4321	0.2732	1600	1.0699
5.0207	0.3415	2000	1.0498
3.4903	0.4098	2400	1.0308
4.3367	0.4781	2800	1.0168
3.786	0.5464	3200	1.0016
3.9173	0.6148	3600	0.9910
3.6045	0.6831	4000	0.9823
3.7615	0.7514	4400	0.9765
3.9633	0.8197	4800	0.9741
3.7323	0.8880	5200	0.9732

Framework versions

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