See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/codegemma-2b
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - e3cc843ef0986df4_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/e3cc843ef0986df4_train_data.json
  type:
    field_input: Complex_Cot
    field_instruction: Question
    field_output: Response
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/ce898271-e75f-469a-a759-eb04d5f4a42e
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
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: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2040
micro_batch_size: 4
mlflow_experiment_name: /tmp/e3cc843ef0986df4_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
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: 0eead81e-cd80-4227-841f-91eba396d2cc
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 0eead81e-cd80-4227-841f-91eba396d2cc
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

ce898271-e75f-469a-a759-eb04d5f4a42e

This model is a fine-tuned version of unsloth/codegemma-2b on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.9249

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: 8
• total_train_batch_size: 32
• 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: 1460

Training results

Training Loss	Epoch	Step	Validation Loss
1.6633	0.0014	1	1.5823
1.1351	0.1370	100	1.1335
1.1085	0.2741	200	1.0738
0.9766	0.4111	300	1.0379
0.9988	0.5481	400	1.0121
0.988	0.6852	500	0.9913
0.9833	0.8222	600	0.9750
1.0019	0.9592	700	0.9570
0.8074	1.0963	800	0.9542
1.0179	1.2333	900	0.9464
0.7274	1.3703	1000	0.9393
1.0496	1.5074	1100	0.9326
0.8693	1.6444	1200	0.9281
0.8627	1.7814	1300	0.9259
0.8052	1.9185	1400	0.9249

Framework versions

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