See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_resume_from_checkpoints: false
base_model: echarlaix/tiny-random-PhiForCausalLM
bf16: auto
chat_template: llama3
dataset_prepared_path: null
dataset_processes: 
datasets:
- data_files:
  - df4145ffec439e54_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/df4145ffec439e54_train_data.json
  type:
    field_instruction: text
    field_output: text_ja
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 150
eval_table_size: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 1
gradient_checkpointing: true
group_by_length: false
hub_model_id: error577/56fde327-537c-4d7d-8a47-5553a75fd430
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
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_grad_norm: 1.0
max_steps: null
micro_batch_size: 512
mlflow_experiment_name: /tmp/df4145ffec439e54_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
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: 150
sequence_len: 512
special_tokens:
  pad_token: <|endoftext|>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.5
wandb_entity: null
wandb_mode: online
wandb_name: c30de9d5-9d24-44e1-9b62-deb3229d1fce
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: c30de9d5-9d24-44e1-9b62-deb3229d1fce
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

56fde327-537c-4d7d-8a47-5553a75fd430

This model is a fine-tuned version of echarlaix/tiny-random-PhiForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 6.6156

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: 512
• eval_batch_size: 512
• 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: 30
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss
6.9448	0.0116	1	6.9446
6.6809	1.7442	150	6.6820
6.6655	3.4884	300	6.6633
6.6563	5.2326	450	6.6522
6.649	6.9767	600	6.6462
6.6442	8.7209	750	6.6431
6.6415	10.4651	900	6.6377
6.6357	12.2093	1050	6.6346
6.6334	13.9535	1200	6.6311
6.6309	15.6977	1350	6.6292
6.6386	17.4419	1500	6.6275
6.6273	19.1860	1650	6.6248
6.627	20.9302	1800	6.6227
6.6243	22.6744	1950	6.6207
6.6252	24.4186	2100	6.6195
6.6204	26.1628	2250	6.6187
6.6222	27.9070	2400	6.6182
6.6231	29.6512	2550	6.6176
6.6196	31.3953	2700	6.6171
6.6161	33.1395	2850	6.6168
6.6213	34.8837	3000	6.6165
6.6209	36.6279	3150	6.6162
6.6176	38.3721	3300	6.6160
6.6219	40.1163	3450	6.6158
6.619	41.8605	3600	6.6157
6.6143	43.6047	3750	6.6156
6.6199	45.3488	3900	6.6156
6.6195	47.0930	4050	6.6156
6.6185	48.8372	4200	6.6156

Framework versions

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