See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/SmolLM-360M-Instruct
bf16: auto
chat_template: llama3
dataloader_num_workers: 6
dataset_prepared_path: null
datasets:
- data_files:
  - 824ef92913906b02_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/824ef92913906b02_train_data.json
  type:
    field_input: gt_answer
    field_instruction: question
    field_output: answer
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 50
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: true
hub_model_id: error577/3b560dce-4b9a-4461-bec5-e809c217723b
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: true
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2000
micro_batch_size: 2
mlflow_experiment_name: /tmp/824ef92913906b02_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 4
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: 50
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.02
wandb_entity: null
wandb_mode: online
wandb_name: edbc27c0-2567-4ef1-849f-694fac00cd13
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: edbc27c0-2567-4ef1-849f-694fac00cd13
warmup_steps: 10
weight_decay: 0.01
xformers_attention: null

3b560dce-4b9a-4461-bec5-e809c217723b

This model is a fine-tuned version of unsloth/SmolLM-360M-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3736

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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: 2000

Training results

Training Loss	Epoch	Step	Validation Loss
0.6086	0.0001	1	0.8242
0.6491	0.0069	50	0.5668
0.4904	0.0137	100	0.5020
0.4928	0.0206	150	0.4803
0.3963	0.0275	200	0.4617
0.4291	0.0343	250	0.4514
0.4164	0.0412	300	0.4411
0.4261	0.0481	350	0.4344
0.4734	0.0549	400	0.4273
0.4271	0.0618	450	0.4246
0.3362	0.0687	500	0.4193
0.3781	0.0755	550	0.4136
0.4247	0.0824	600	0.4107
0.3431	0.0893	650	0.4071
0.3674	0.0961	700	0.4019
0.4549	0.1030	750	0.4004
0.4299	0.1098	800	0.3972
0.3398	0.1167	850	0.3938
0.4542	0.1236	900	0.3924
0.301	0.1304	950	0.3908
0.4517	0.1373	1000	0.3881
0.4243	0.1442	1050	0.3865
0.3469	0.1510	1100	0.3853
0.3413	0.1579	1150	0.3834
0.3609	0.1648	1200	0.3815
0.3546	0.1716	1250	0.3813
0.2967	0.1785	1300	0.3794
0.3278	0.1854	1350	0.3783
0.3304	0.1922	1400	0.3782
0.2565	0.1991	1450	0.3774
0.3721	0.2060	1500	0.3764
0.4291	0.2128	1550	0.3754
0.299	0.2197	1600	0.3757
0.4525	0.2266	1650	0.3750
0.3902	0.2334	1700	0.3747
0.3118	0.2403	1750	0.3737
0.3434	0.2472	1800	0.3747
0.3902	0.2540	1850	0.3732
0.3748	0.2609	1900	0.3742
0.3463	0.2678	1950	0.3742
0.3441	0.2746	2000	0.3736

Framework versions

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