See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: NousResearch/CodeLlama-7b-hf
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 77f4d29f468bc2c0_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/77f4d29f468bc2c0_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: response
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 0.001
eval_max_new_tokens: 128
eval_steps: 40
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/ee29425a-d835-495b-8fb5-e11141a51ea1
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 100
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
micro_batch_size: 32
mlflow_experiment_name: /tmp/77f4d29f468bc2c0_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
s2_attention: null
sample_packing: false
save_steps: 40
saves_per_epoch: 0
sequence_len: 512
special_tokens:
  pad_token: </s>
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: 48c3c8d8-6645-4a88-adc5-70cd656a2562
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 48c3c8d8-6645-4a88-adc5-70cd656a2562
warmup_ratio: 0.05
weight_decay: 0.0
xformers_attention: null

ee29425a-d835-495b-8fb5-e11141a51ea1

This model is a fine-tuned version of NousResearch/CodeLlama-7b-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6540

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: 0.0003
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_bnb_8bit 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: 218
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0014	1	2.6242
No log	0.0571	40	2.2860
No log	0.1142	80	1.9066
4.9536	0.1713	120	1.7338
4.9536	0.2284	160	1.6253
3.6406	0.2855	200	1.5644
3.6406	0.3426	240	1.4934
3.6406	0.3997	280	1.4442
3.1356	0.4568	320	1.3714
3.1356	0.5139	360	1.3406
2.8521	0.5710	400	1.3140
2.8521	0.6281	440	1.2290
2.8521	0.6852	480	1.1974
2.7573	0.7423	520	1.1788
2.7573	0.7994	560	1.1294
2.3985	0.8565	600	1.1246
2.3985	0.9136	640	1.0637
2.3985	0.9707	680	1.0167
2.2745	1.0278	720	1.0103
2.2745	1.0849	760	0.9873
1.8141	1.1420	800	0.9797
1.8141	1.1991	840	0.9597
1.8141	1.2562	880	0.9458
1.5585	1.3133	920	0.9204
1.5585	1.3704	960	0.9172
1.5745	1.4276	1000	0.9020
1.5745	1.4847	1040	0.8774
1.5745	1.5418	1080	0.8412
1.3623	1.5989	1120	0.8341
1.3623	1.6560	1160	0.8323
1.3262	1.7131	1200	0.7996
1.3262	1.7702	1240	0.8067
1.3262	1.8273	1280	0.7752
1.2514	1.8844	1320	0.7176
1.2514	1.9415	1360	0.7170
1.2277	1.9986	1400	0.7039
1.2277	2.0557	1440	0.7033
1.2277	2.1128	1480	0.7025
0.68	2.1699	1520	0.7079
0.68	2.2270	1560	0.7214
0.809	2.2841	1600	0.6897
0.809	2.3412	1640	0.7047
0.809	2.3983	1680	0.7042
0.7931	2.4554	1720	0.6792
0.7931	2.5125	1760	0.6777
0.6957	2.5696	1800	0.6809
0.6957	2.6267	1840	0.6466
0.6957	2.6838	1880	0.6705
0.7611	2.7409	1920	0.6549
0.7611	2.7980	1960	0.6540

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.1