See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: hf-tiny-model-private/tiny-random-OPTForCausalLM
bf16: true
chat_template: llama3
dataloader_num_workers: 6
dataset_prepared_path: null
datasets:
- data_files:
  - 9742221b6e81649f_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/9742221b6e81649f_train_data.json
  type:
    field_input: language
    field_instruction: company
    field_output: sentence
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping:
  metric: eval_loss
  mode: min
  patience: 3
eval_max_new_tokens: 128
eval_steps: 200
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: true
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: true
hub_model_id: error577/942cfd35-534e-4ab7-9dd2-335ce292d7c2
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3500
micro_batch_size: 16
mlflow_experiment_name: /tmp/9742221b6e81649f_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
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: 200
sequence_len: 64
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: 1be7f53b-7c51-4d01-9396-e086bf3f09dd
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 1be7f53b-7c51-4d01-9396-e086bf3f09dd
warmup_steps: 10
weight_decay: 0.01
xformers_attention: null

942cfd35-534e-4ab7-9dd2-335ce292d7c2

This model is a fine-tuned version of hf-tiny-model-private/tiny-random-OPTForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 6.9015

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: 3500

Training results

Training Loss	Epoch	Step	Validation Loss
27.7344	0.0025	1	6.9353
27.4324	0.5057	200	6.9087
27.6605	1.0114	400	6.9080
27.6448	1.5171	600	6.9054
27.6474	2.0228	800	6.9055
27.6403	2.5284	1000	6.9049
27.6381	3.0341	1200	6.9040
27.6395	3.5398	1400	6.9025
27.6276	4.0455	1600	6.9028
27.6327	4.5512	1800	6.9028
27.6274	5.0569	2000	6.9020
27.6326	5.5626	2200	6.9019
27.6167	6.0683	2400	6.9010
27.645	6.5740	2600	6.9005
27.6261	7.0796	2800	6.9017
27.6214	7.5853	3000	6.9015
27.6217	8.0910	3200	6.9016
27.6327	8.5967	3400	6.9015

Framework versions

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