See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: echarlaix/tiny-random-PhiForCausalLM
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 5f27bc50ea77f27e_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/5f27bc50ea77f27e_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: output
    format: '{instruction} {input}'
    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: 100
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/ea51a844-1173-4395-b5f2-5ab8cd5edc35
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: 2520
micro_batch_size: 4
mlflow_experiment_name: /tmp/5f27bc50ea77f27e_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
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: 100
sequence_len: 1024
special_tokens:
  pad_token: <|endoftext|>
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: 21d80ea0-1a50-4729-b510-a987e272f042
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 21d80ea0-1a50-4729-b510-a987e272f042
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

ea51a844-1173-4395-b5f2-5ab8cd5edc35

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: 6.6190

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
• gradient_accumulation_steps: 8
• total_train_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: 10
• training_steps: 1889

Training results

Training Loss	Epoch	Step	Validation Loss
6.9462	0.0011	1	6.9452
6.8211	0.1059	100	6.8012
6.7556	0.2118	200	6.7375
6.7281	0.3178	300	6.7051
6.6707	0.4237	400	6.6791
6.6773	0.5296	500	6.6604
6.7186	0.6355	600	6.6482
6.6813	0.7414	700	6.6399
6.6778	0.8473	800	6.6341
6.6404	0.9533	900	6.6297
6.4898	1.0592	1000	6.6267
6.7481	1.1651	1100	6.6241
6.5533	1.2710	1200	6.6226
7.0404	1.3769	1300	6.6214
7.7962	1.4829	1400	6.6202
6.7513	1.5888	1500	6.6196
6.6431	1.6947	1600	6.6192
6.2248	1.8006	1700	6.6191
6.1459	1.9065	1800	6.6190

Framework versions

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