See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: princeton-nlp/Sheared-LLaMA-1.3B
bf16: true
chat_template: llama3
cosine_min_lr_ratio: 0.3
dataset_prepared_path: null
datasets:
- data_files:
  - 9111ab42dd82442c_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/9111ab42dd82442c_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
deepspeed: null
early_stopping_patience: 4
eval_max_new_tokens: 128
eval_steps: 200
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/ff233829-a82a-4192-8f8d-1f8a03a1a77d
hub_repo: null
hub_strategy: checkpoint
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: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 6052
micro_batch_size: 4
mlflow_experiment_name: /tmp/9111ab42dd82442c_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_torch
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 200
sequence_len: 2048
special_tokens:
  pad_token: </s>
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.045059252917586626
wandb_entity: null
wandb_mode: online
wandb_name: 71ac6567-ec89-44ef-90cf-3963c0514683
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 71ac6567-ec89-44ef-90cf-3963c0514683
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

ff233829-a82a-4192-8f8d-1f8a03a1a77d

This model is a fine-tuned version of princeton-nlp/Sheared-LLaMA-1.3B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.6631

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: 4
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH 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: 6052

Training results

Training Loss	Epoch	Step	Validation Loss
2.8134	0.0002	1	2.7262
2.0828	0.0302	200	2.1091
2.1855	0.0604	400	2.0203
1.8754	0.0906	600	1.9702
2.0078	0.1208	800	1.9297
1.9068	0.1510	1000	1.9031
2.1842	0.1812	1200	1.8784
1.9692	0.2114	1400	1.8576
1.5986	0.2416	1600	1.8413
1.8089	0.2718	1800	1.8265
2.0088	0.3020	2000	1.8140
1.6978	0.3322	2200	1.8014
1.6894	0.3624	2400	1.7903
1.3886	0.3926	2600	1.7749
1.7118	0.4228	2800	1.7639
1.7178	0.4530	3000	1.7588
1.7374	0.4832	3200	1.7460
1.7989	0.5134	3400	1.7356
1.7476	0.5436	3600	1.7266
2.116	0.5738	3800	1.7193
1.7626	0.6040	4000	1.7119
1.83	0.6342	4200	1.7047
1.8368	0.6644	4400	1.6997
1.3722	0.6945	4600	1.6924
1.9036	0.7247	4800	1.6884
1.6224	0.7549	5000	1.6830
1.5629	0.7851	5200	1.6776
1.9096	0.8153	5400	1.6730
1.8817	0.8455	5600	1.6696
1.7826	0.8757	5800	1.6655
1.545	0.9059	6000	1.6631

Framework versions

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