See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: true
base_model: NousResearch/Nous-Hermes-2-SOLAR-10.7B
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 25696e83f649cdb7_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/25696e83f649cdb7_train_data.json
  type:
    field_input: context
    field_instruction: QuestionText
    field_output: question
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
do_eval: true
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 50
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 2
gradient_checkpointing: false
group_by_length: true
hub_model_id: lesso15/e4ec0f07-38c0-4df9-a9a1-bbfeec2abbe6
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.000215
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 10
lora_alpha: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 500
micro_batch_size: 4
mlflow_experiment_name: /ephemeral/tmp/G.O.D/25696e83f649cdb7_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: 50
saves_per_epoch: null
seed: 150
sequence_len: 512
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: c8c2d8aa-92bc-4121-b828-82c8dec7071c
wandb_project: 15a
wandb_run: your_name
wandb_runid: c8c2d8aa-92bc-4121-b828-82c8dec7071c
warmup_steps: 50
weight_decay: 0.0
xformers_attention: null

e4ec0f07-38c0-4df9-a9a1-bbfeec2abbe6

This model is a fine-tuned version of NousResearch/Nous-Hermes-2-SOLAR-10.7B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0574

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: 0.000215
• train_batch_size: 4
• eval_batch_size: 4
• seed: 150
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• 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: 50
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0003	1	3.7715
0.166	0.0135	50	0.2286
0.164	0.0270	100	0.1427
0.0903	0.0405	150	0.1290
0.0872	0.0540	200	0.1427
0.0878	0.0675	250	0.0953
0.0754	0.0810	300	0.0729
0.0695	0.0945	350	0.0659
0.0499	0.1080	400	0.0610
0.0647	0.1215	450	0.0590
0.0561	0.1350	500	0.0574

Framework versions

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