See axolotl config

axolotl version: 0.4.1

adapter: qlora
auto_resume_from_checkpoints: true
base_model: facebook/opt-350m
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 9b3a5919e996b43f_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/9b3a5919e996b43f_train_data.json
  type:
    field_input: thinking
    field_instruction: prompt
    field_output: answer
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
evals_per_epoch: null
flash_attention: false
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: true
hub_model_id: error577/bab80e07-cf92-4525-abe6-bf0d64212509
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 10
lora_alpha: 128
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: null
micro_batch_size: 2
mlflow_experiment_name: /tmp/9b3a5919e996b43f_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_torch_4bit
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: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: ef00d239-571d-44fb-ae4b-a020df7f09b1
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: ef00d239-571d-44fb-ae4b-a020df7f09b1
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

bab80e07-cf92-4525-abe6-bf0d64212509

This model is a fine-tuned version of facebook/opt-350m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6809

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_4BIT 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: 30
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0005	1	1.7855
5.1691	0.0537	100	1.2518
4.7876	0.1075	200	1.1656
4.3157	0.1612	300	1.0780
4.2875	0.2150	400	1.0325
4.4765	0.2687	500	0.9903
4.1246	0.3225	600	0.9688
3.8197	0.3762	700	0.9726
3.8941	0.4300	800	0.9635
4.1191	0.4837	900	0.9212
3.6611	0.5375	1000	0.9328
3.5027	0.5912	1100	0.8896
3.4397	0.6450	1200	0.9109
3.5156	0.6987	1300	0.8761
3.7601	0.7525	1400	0.8751
3.4886	0.8062	1500	0.8606
3.1389	0.8600	1600	0.8488
3.2549	0.9137	1700	0.8776
3.4852	0.9675	1800	0.8418
3.3517	1.0212	1900	0.8019
3.1547	1.0750	2000	0.8292
3.1288	1.1287	2100	0.7916
2.9624	1.1825	2200	0.8047
3.1523	1.2362	2300	0.7841
3.01	1.2900	2400	0.7736
3.2158	1.3437	2500	0.7726
3.0904	1.3975	2600	0.7758
2.9655	1.4512	2700	0.7682
3.1154	1.5050	2800	0.7612
3.199	1.5587	2900	0.7508
3.1015	1.6125	3000	0.7463
2.9626	1.6662	3100	0.7425
3.0086	1.7200	3200	0.7358
2.9609	1.7737	3300	0.7343
2.7884	1.8275	3400	0.7286
2.9546	1.8812	3500	0.7227
2.8298	1.9350	3600	0.7203
2.8598	1.9887	3700	0.7133
2.5154	2.0425	3800	0.7121
2.6962	2.0962	3900	0.7085
2.7309	2.1500	4000	0.7024
2.6869	2.2037	4100	0.7035
2.7636	2.2575	4200	0.6971
2.7232	2.3112	4300	0.6901
2.8095	2.3650	4400	0.6923
2.725	2.4187	4500	0.6882
2.793	2.4725	4600	0.6892
2.6342	2.5262	4700	0.6852
2.8126	2.5800	4800	0.6832
2.4884	2.6337	4900	0.6812
2.6875	2.6874	5000	0.6800
2.7025	2.7412	5100	0.6811
2.6606	2.7949	5200	0.6808
2.5986	2.8487	5300	0.6809

Framework versions

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