See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: unsloth/SmolLM2-360M-Instruct
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 30b469a6bf8051b9_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/30b469a6bf8051b9_train_data.json
  type:
    field_input: image
    field_instruction: options
    field_output: question
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 0.0001
eval_max_new_tokens: 128
eval_steps: 133
eval_strategy: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/5a1bfcde-5bde-48b2-943e-9961981f45dd
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0004
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 133
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_steps: 
micro_batch_size: 32
mlflow_experiment_name: /tmp/30b469a6bf8051b9_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 100
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: /workspace/hub_repo/last-checkpoint
s2_attention: null
sample_packing: false
save_steps: 133
saves_per_epoch: 0
sequence_len: 512
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: 
wandb_name: cca84863-925c-4f9b-bfe6-f38dbb6a1908
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: cca84863-925c-4f9b-bfe6-f38dbb6a1908
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

5a1bfcde-5bde-48b2-943e-9961981f45dd

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

• Loss: 1.4550

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: 0.0004
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: 100
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0007	1	2.8914
2.055	0.1080	160	1.7022
1.6504	0.2160	320	1.6316
1.6044	0.3240	480	1.5953
1.5876	0.4320	640	1.5720
1.5508	0.5400	800	1.5623
1.5527	0.6480	960	1.5444
1.5412	0.7560	1120	1.5334
1.5165	0.8640	1280	1.5204
1.5192	0.9720	1440	1.5123
1.4741	1.0800	1600	1.5085
1.452	1.1880	1760	1.4991
1.4549	1.2960	1920	1.4922
1.452	1.4040	2080	1.4876
1.4404	1.5120	2240	1.4804
1.4394	1.6200	2400	1.4755
1.4422	1.7280	2560	1.4685
1.4255	1.8360	2720	1.4650
1.4217	1.9440	2880	1.4616
1.3806	2.0520	3040	1.4647
1.3477	2.1600	3200	1.4654
1.3519	2.2680	3360	1.4577
1.3502	2.3760	3520	1.4573
1.3556	2.4840	3680	1.4490
1.3568	2.5920	3840	1.4474
1.3564	2.7000	4000	1.4454
1.3425	2.8080	4160	1.4430
1.3361	2.9160	4320	1.4392
1.3337	3.0240	4480	1.4506
1.2435	3.1320	4640	1.4572
1.2629	3.2400	4800	1.4550

Framework versions

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