See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/SmolLM2-135M
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - a0682f06fa4fb877_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/a0682f06fa4fb877_train_data.json
  type:
    field_instruction: instruction
    field_output: positive_sample
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 1
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: error577/d09828c9-a63a-4168-9212-271729f5e2cf
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 0.0001
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_steps: 1000
micro_batch_size: 2
mlflow_experiment_name: /tmp/a0682f06fa4fb877_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 20
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
saves_per_epoch: 1
sequence_len: 4096
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: 981e1948-88b4-489c-9ec2-583892ddfa73
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 981e1948-88b4-489c-9ec2-583892ddfa73
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

d09828c9-a63a-4168-9212-271729f5e2cf

This model is a fine-tuned version of unsloth/SmolLM2-135M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.8314

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
• gradient_accumulation_steps: 8
• 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: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
2.9265	0.0056	1	3.1834
2.6463	0.2797	50	2.9924
2.7358	0.5594	100	2.9361
2.7248	0.8392	150	2.9165
1.7698	1.1203	200	2.9056
2.663	1.4	250	2.8940
2.577	1.6797	300	2.8816
3.151	1.9594	350	2.8750
2.8115	2.2406	400	2.8675
3.0385	2.5203	450	2.8571
2.8689	2.8	500	2.8508
2.7139	3.0811	550	2.8480
3.3137	3.3608	600	2.8435
2.0524	3.6406	650	2.8411
3.0692	3.9203	700	2.8377
2.8281	4.2014	750	2.8348
2.7376	4.4811	800	2.8323
2.9253	4.7608	850	2.8327
3.1771	5.0420	900	2.8334
1.8451	5.3217	950	2.8321
1.9423	5.6014	1000	2.8314

Framework versions

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