See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Llama-3.2-1B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 62826f43b9114472_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/62826f43b9114472_train_data.json
  type:
    field_input: seed
    field_instruction: instruction
    field_output: response
    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: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/f2b3423d-d7bb-4141-ba74-4495c4913457
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: 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
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2520
micro_batch_size: 4
mlflow_experiment_name: /tmp/62826f43b9114472_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: 2048
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: 69f00a31-a03d-4a9b-b858-6c79d6a94257
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 69f00a31-a03d-4a9b-b858-6c79d6a94257
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

f2b3423d-d7bb-4141-ba74-4495c4913457

This model is a fine-tuned version of unsloth/Llama-3.2-1B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5037

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: 2520

Training results

Training Loss	Epoch	Step	Validation Loss
0.8121	0.0007	1	0.8412
0.5168	0.0669	100	0.5506
0.5156	0.1337	200	0.5424
0.6124	0.2006	300	0.5392
0.5411	0.2674	400	0.5362
0.5146	0.3343	500	0.5332
0.5125	0.4011	600	0.5286
0.4581	0.4680	700	0.5261
0.5059	0.5348	800	0.5248
0.5774	0.6017	900	0.5196
0.5751	0.6685	1000	0.5167
0.4728	0.7354	1100	0.5132
0.5323	0.8022	1200	0.5086
0.4639	0.8691	1300	0.5043
0.5134	0.9359	1400	0.5020
0.4485	1.0032	1500	0.4993
0.4138	1.0700	1600	0.5027
0.4754	1.1369	1700	0.5037

Framework versions

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