See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2.5-0.5B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - f2f0b8a70078fb6f_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/f2f0b8a70078fb6f_train_data.json
  type:
    field_input: Category
    field_instruction: Description
    field_output: Product Name
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,4
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 33
eval_table_size: null
flash_attention: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/e3dccf45-cefa-4f2a-a0fb-666c8f060ecb
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: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2176.0
micro_batch_size: 4
mlflow_experiment_name: /tmp/f2f0b8a70078fb6f_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: 33
sequence_len: 1024
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: a301c3a8-0dab-457d-9ed0-613fdf7a26f0
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: a301c3a8-0dab-457d-9ed0-613fdf7a26f0
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

e3dccf45-cefa-4f2a-a0fb-666c8f060ecb

This model is a fine-tuned version of unsloth/Qwen2.5-0.5B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8577

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
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• total_eval_batch_size: 16
• 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: 494

Training results

Training Loss	Epoch	Step	Validation Loss
4.5112	0.0040	1	4.5996
2.984	0.1336	33	2.9557
2.5622	0.2672	66	2.5605
2.2693	0.4008	99	2.2670
2.0048	0.5344	132	2.0253
1.6808	0.6680	165	1.7832
1.5331	0.8016	198	1.5485
1.5456	0.9352	231	1.3710
0.9004	1.0688	264	1.2195
0.9746	1.2024	297	1.1203
0.9145	1.3360	330	1.0257
0.9218	1.4696	363	0.9484
0.8369	1.6032	396	0.8949
0.7118	1.7368	429	0.8671
0.7696	1.8704	462	0.8577

Framework versions

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