See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: NousResearch/Hermes-2-Theta-Llama-3-8B
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - aef18cd6aa739768_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/aef18cd6aa739768_train_data.json
  type:
    field_instruction: question
    field_output: reference_answer
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
ema_decay: 0.9992
eval_max_new_tokens: 128
eval_table_size: null
evals_per_epoch: 4
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: JoshMe1/c187d5a0-b524-492e-ae22-f5ad07a9ed9a
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0001
load_in_4bit: false
load_in_8bit: true
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
lr_scheduler: cosine
max_steps: 122
micro_batch_size: 4
mlflow_experiment_name: /tmp/aef18cd6aa739768_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
saves_per_epoch: 4
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
use_ema: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: befca2b0-af17-45b9-a5aa-8213942874a0
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: befca2b0-af17-45b9-a5aa-8213942874a0
warmup_steps: 10
weight_decay: 0.01
xformers_attention: true

c187d5a0-b524-492e-ae22-f5ad07a9ed9a

This model is a fine-tuned version of NousResearch/Hermes-2-Theta-Llama-3-8B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0588

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 122

Training results

Training Loss	Epoch	Step	Validation Loss
3.4806	0.0017	1	3.0263
1.3969	0.0275	16	1.2720
0.9667	0.0550	32	0.7062
0.5098	0.0824	48	0.3667
0.1669	0.1099	64	0.2007
0.0077	0.1374	80	0.1537
0.2602	0.1649	96	0.0902
0.0094	0.1924	112	0.0588

Framework versions

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