See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: HuggingFaceH4/tiny-random-LlamaForCausalLM
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 24a19a7ab43df05e_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/24a19a7ab43df05e_train_data.json
  type:
    field_instruction: role
    field_output: text
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: romainnn/0986a89f-885c-4522-a4b0-1673f1d20546
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
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
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 1980
micro_batch_size: 4
mlflow_experiment_name: /tmp/24a19a7ab43df05e_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
save_steps: 100
sequence_len: 2048
special_tokens:
  pad_token: </s>
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.0370441714700609
wandb_entity: null
wandb_mode: online
wandb_name: 93576204-0c9b-4c0d-8f63-79add5c5fd0b
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 93576204-0c9b-4c0d-8f63-79add5c5fd0b
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

0986a89f-885c-4522-a4b0-1673f1d20546

This model is a fine-tuned version of HuggingFaceH4/tiny-random-LlamaForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 10.3498

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: 1980

Training results

Training Loss	Epoch	Step	Validation Loss
10.3776	0.0002	1	10.3774
10.3616	0.0246	100	10.3609
10.3572	0.0492	200	10.3583
10.3564	0.0739	300	10.3559
10.3576	0.0985	400	10.3549
10.3541	0.1231	500	10.3537
10.3508	0.1477	600	10.3525
10.3496	0.1723	700	10.3519
10.3489	0.1970	800	10.3514
10.3555	0.2216	900	10.3510
10.3512	0.2462	1000	10.3507
10.3474	0.2708	1100	10.3505
10.3529	0.2954	1200	10.3503
10.3528	0.3201	1300	10.3501
10.3526	0.3447	1400	10.3500
10.3521	0.3693	1500	10.3499
10.3542	0.3939	1600	10.3498
10.349	0.4185	1700	10.3498
10.3473	0.4432	1800	10.3498
10.3506	0.4678	1900	10.3498

Framework versions

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