See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/tinyllama
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - e029f217fa002728_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/e029f217fa002728_train_data.json
  type:
    field_input: overview
    field_instruction: raw_text
    field_output: clean_text
    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/b44edac3-df02-44f0-a749-5b89fdf83692
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: 2036
micro_batch_size: 4
mlflow_experiment_name: /tmp/e029f217fa002728_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: f214cfe8-8866-498c-ad88-a995718d9d2d
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: f214cfe8-8866-498c-ad88-a995718d9d2d
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

b44edac3-df02-44f0-a749-5b89fdf83692

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

• Loss: 0.1427

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: 2036

Training results

Training Loss	Epoch	Step	Validation Loss
4.4824	0.0009	1	4.7982
0.1726	0.0897	100	0.1894
0.2216	0.1794	200	0.1676
0.032	0.2690	300	0.1616
0.0271	0.3587	400	0.1539
0.3435	0.4484	500	0.1514
0.0376	0.5381	600	0.1493
0.1738	0.6277	700	0.1479
0.022	0.7174	800	0.1471
0.2307	0.8071	900	0.1466
0.0315	0.8968	1000	0.1450
0.1236	0.9864	1100	0.1442
0.1171	1.0761	1200	0.1444
0.037	1.1658	1300	0.1441
0.205	1.2555	1400	0.1438
0.0886	1.3451	1500	0.1434
0.3406	1.4348	1600	0.1431
0.1887	1.5245	1700	0.1430
0.032	1.6142	1800	0.1428
0.3551	1.7038	1900	0.1427
0.0924	1.7935	2000	0.1427

Framework versions

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