See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: EleutherAI/pythia-70m-deduped
bf16: auto
dataset_prepared_path: null
datasets:
- data_files:
  - e1681b7389f97e66_train_data.json
  ds_type: json
  format: custom
  path: e1681b7389f97e66_train_data.json
  type:
    field: null
    field_input: null
    field_instruction: Question
    field_output: Answer
    field_system: null
    format: null
    no_input_format: null
    system_format: '{system}'
    system_prompt: ''
early_stopping_patience: null
evals_per_epoch: 3
gradient_accumulation_steps: 1
group_by_length: false
hub_model_id: taopanda/5e5817db-f8a6-4e92-b422-85db307780cd
learning_rate: 1.0e-05
load_in_8bit: true
local_rank: null
logging_steps: 1
lora_alpha: 32
lora_dropout: 0.05
lora_fan_in_fan_out: true
lora_model_dir: null
lora_r: 16
lora_target_linear: null
lora_target_modules:
- query_key_value
micro_batch_size: 4
num_epochs: 1
output_dir: ./outputs/lora-alpaca-pythia/taopanda-2_a1e79dad-bd95-489a-8ac5-0c8fa2c2d2ff
resume_from_checkpoint: null
seed: 41249
sequence_len: 512
special_tokens:
  pad_token: <|endoftext|>
tf32: true
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: fatcat87-taopanda
wandb_log_model: null
wandb_mode: online
wandb_name: taopanda-2_a1e79dad-bd95-489a-8ac5-0c8fa2c2d2ff
wandb_project: subnet56
wandb_runid: taopanda-2_a1e79dad-bd95-489a-8ac5-0c8fa2c2d2ff
wandb_watch: null
weight_decay: 0.1

5e5817db-f8a6-4e92-b422-85db307780cd

This model is a fine-tuned version of EleutherAI/pythia-70m-deduped on the None dataset. It achieves the following results on the evaluation set:

• Loss: 53.2413

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 41249
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 16
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 4
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
60.4809	0.0065	1	53.4904
52.756	0.3333	51	53.3550
56.5209	0.6667	102	53.3138
54.042	1.0	153	53.2413

Framework versions

• PEFT 0.11.1
• Transformers 4.42.3
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1