See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_resume_from_checkpoints: false
base_model: unsloth/SmolLM2-1.7B-Instruct
bf16: auto
chat_template: llama3
dataset_prepared_path: null
dataset_processes: 6
datasets:
- data_files:
  - c19ea90a9b1e15e2_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/c19ea90a9b1e15e2_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: output
    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: 200
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: error577/d6d23524-6634-4e56-a2ee-8c47d4ce637b
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.1
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: null
micro_batch_size: 2
mlflow_experiment_name: /tmp/c19ea90a9b1e15e2_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: 200
sequence_len: 256
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: dff7995c-07d8-4bb9-994f-f1206b76fb84
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: dff7995c-07d8-4bb9-994f-f1206b76fb84
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

d6d23524-6634-4e56-a2ee-8c47d4ce637b

This model is a fine-tuned version of unsloth/SmolLM2-1.7B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3850

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: 30
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
1.0223	0.0001	1	1.1713
0.5253	0.0239	200	0.5158
0.6398	0.0478	400	0.4853
0.7508	0.0718	600	0.4676
0.4842	0.0957	800	0.4558
0.5774	0.1196	1000	0.4508
0.5879	0.1435	1200	0.4474
0.611	0.1674	1400	0.4384
0.2718	0.1914	1600	0.4334
0.5524	0.2153	1800	0.4317
0.5781	0.2392	2000	0.4269
0.24	0.2631	2200	0.4222
0.5552	0.2871	2400	0.4226
0.5357	0.3110	2600	0.4221
0.5229	0.3349	2800	0.4214
0.4939	0.3588	3000	0.4166
0.4376	0.3827	3200	0.4142
0.4528	0.4067	3400	0.4144
0.459	0.4306	3600	0.4093
0.4331	0.4545	3800	0.4098
0.2921	0.4784	4000	0.4078
0.1995	0.5023	4200	0.4049
0.3215	0.5263	4400	0.4018
0.5111	0.5502	4600	0.4014
0.2931	0.5741	4800	0.3999
0.3265	0.5980	5000	0.4022
0.6211	0.6220	5200	0.4016
0.6088	0.6459	5400	0.3957
0.3732	0.6698	5600	0.3977
0.4201	0.6937	5800	0.3944
0.5265	0.7176	6000	0.3962
0.4367	0.7416	6200	0.3933
0.4757	0.7655	6400	0.3908
0.2699	0.7894	6600	0.3898
0.4476	0.8133	6800	0.3863
0.4146	0.8372	7000	0.3869
0.5522	0.8612	7200	0.3843
0.4478	0.8851	7400	0.3833
0.4299	0.9090	7600	0.3847
0.4298	0.9329	7800	0.3858
0.4177	0.9569	8000	0.3850

Framework versions

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