See axolotl config

axolotl version: 0.5.2

adapter: lora
auto_find_batch_size: true
base_model: unsloth/SmolLM2-360M
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - eaee6ab7fe1ea1c6_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/eaee6ab7fe1ea1c6_train_data.json
  type:
    field_input: text
    field_instruction: instruction
    field_output: Resume_test
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: /workspace/axolotl/configs/deepspeed_stage2.json
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_sample_packing: false
eval_steps: 10
eval_strategy: steps
eval_table_size: null
flash_attention: true
fp16: false
gpu_memory_limit: 80GiB
gradient_accumulation_steps: 4
gradient_checkpointing: true
greater_is_better: false
group_by_length: true
hub_model_id: PhoenixB/6d50d116-df0c-4b2c-a38b-84cd30213fcd
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 1e-4
liger_fused_linear_cross_entropy: true
liger_glu_activation: true
liger_layer_norm: true
liger_rms_norm: true
liger_rope: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 5
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lr_scheduler: cosine
max_steps: 10000
metric_for_best_model: loss
micro_batch_size: 2
mlflow_experiment_name: /tmp/eaee6ab7fe1ea1c6_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_torch_fused
output_dir: miner_id_24
pad_to_sequence_len: true
plugins:
- axolotl.integrations.liger.LigerPlugin
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 20
sequence_len: 8192
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 9aacdce9-8545-4665-b64f-dee5a9f66890
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 9aacdce9-8545-4665-b64f-dee5a9f66890
warmup_steps: 20
weight_decay: 0.0

6d50d116-df0c-4b2c-a38b-84cd30213fcd

This model is a fine-tuned version of unsloth/SmolLM2-360M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0092

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: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 20
• training_steps: 5608

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0005	1	0.0957
0.0666	0.0053	10	0.0951
0.0718	0.0107	20	0.0916
0.0955	0.0160	30	0.0762
0.0435	0.0214	40	0.0620
0.0716	0.0267	50	0.0516
0.0257	0.0321	60	0.0427
0.029	0.0374	70	0.0380
0.0439	0.0428	80	0.0334
0.0173	0.0481	90	0.0302
0.036	0.0535	100	0.0272
0.0125	0.0588	110	0.0244
0.0165	0.0642	120	0.0224
0.0889	0.0695	130	0.0196
0.008	0.0749	140	0.0176
0.0128	0.0802	150	0.0159
0.0053	0.0856	160	0.0144
0.0066	0.0909	170	0.0133
0.0164	0.0963	180	0.0130
0.0036	0.1016	190	0.0124
0.0043	0.1070	200	0.0121
0.0038	0.1123	210	0.0115
0.0034	0.1177	220	0.0108
0.0052	0.1230	230	0.0102
0.0029	0.1284	240	0.0104
0.0036	0.1337	250	0.0103
0.0025	0.1391	260	0.0098
0.0016	0.1444	270	0.0094
0.0042	0.1498	280	0.0093
0.0022	0.1551	290	0.0094
0.0023	0.1605	300	0.0092
0.0024	0.1658	310	0.0095
0.0026	0.1712	320	0.0096
0.0051	0.1765	330	0.0092

Framework versions

• PEFT 0.13.2
• Transformers 4.46.3
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3