See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/SmolLM2-1.7B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - e2ca8fa901ae9dd7_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/e2ca8fa901ae9dd7_train_data.json
  type:
    field_instruction: prompt
    field_output: target
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 30
eval_max_new_tokens: 128
eval_steps: 50
eval_table_size: null
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 16
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/2254aed2-2103-46f0-beda-20abbe00dbdb
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: true
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
micro_batch_size: 4
mlflow_experiment_name: /tmp/e2ca8fa901ae9dd7_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
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: 100
sequence_len: 1024
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: e2815d2a-97ba-4bff-aa3c-18e11ee955e6
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: e2815d2a-97ba-4bff-aa3c-18e11ee955e6
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

2254aed2-2103-46f0-beda-20abbe00dbdb

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

• Loss: nan

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• 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: 10
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.0	0.0003	1	nan
0.0	0.0136	50	nan
0.0	0.0272	100	nan
0.0	0.0408	150	nan
0.0	0.0544	200	nan
0.0	0.0680	250	nan
0.0	0.0816	300	nan
0.0	0.0952	350	nan
0.0	0.1088	400	nan
0.0	0.1224	450	nan
0.0	0.1359	500	nan
0.0	0.1495	550	nan
0.0	0.1631	600	nan
0.0	0.1767	650	nan
0.0	0.1903	700	nan
0.0	0.2039	750	nan
0.0	0.2175	800	nan
0.0	0.2311	850	nan
0.0	0.2447	900	nan
0.0	0.2583	950	nan
0.0	0.2719	1000	nan
0.0	0.2855	1050	nan
0.0	0.2991	1100	nan
0.0	0.3127	1150	nan
0.0	0.3263	1200	nan
0.0	0.3399	1250	nan
0.0	0.3535	1300	nan
0.0	0.3671	1350	nan
0.0	0.3807	1400	nan
0.0	0.3942	1450	nan
0.0	0.4078	1500	nan
0.0	0.4214	1550	nan
0.0	0.4350	1600	nan

Framework versions

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