See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Qwen/Qwen2.5-3B-Instruct
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - cd2ead58134fa71e_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/cd2ead58134fa71e_train_data.json
  type:
    field_instruction: source
    field_output: target
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device_map: auto
do_eval: true
early_stopping_patience: null
eval_batch_size: 2
eval_max_new_tokens: 128
eval_steps: null
eval_table_size: null
evals_per_epoch: 4
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: true
hub_model_id: prxy5604/59a2f3bc-5f85-4a86-842d-2e35f769598d
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0001
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 5
lora_alpha: 128
lora_dropout: 0.05
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_memory:
  0: 75GB
max_steps: 200
micro_batch_size: 4
mlflow_experiment_name: /tmp/cd2ead58134fa71e_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
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: null
saves_per_epoch: null
sequence_len: 1024
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 1559d98e-1fed-41c2-8ba2-72b6489efca4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 1559d98e-1fed-41c2-8ba2-72b6489efca4
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

59a2f3bc-5f85-4a86-842d-2e35f769598d

This model is a fine-tuned version of Qwen/Qwen2.5-3B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8736

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_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: 95

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0317	1	3.3631
3.1644	0.2540	8	2.9340
2.5505	0.5079	16	2.1190
2.2072	0.7619	24	1.9973
1.9066	1.0159	32	1.8940
1.5955	1.2698	40	1.8810
1.6197	1.5238	48	1.8771
1.6287	1.7778	56	1.8360
1.579	2.0317	64	1.7921
1.2425	2.2857	72	1.8616
1.3189	2.5397	80	1.8305
1.1198	2.7937	88	1.8736

Framework versions

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