adapter: lora
base_model: samoline/651e2cac-f73e-483e-9e71-50976a75fcaa
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 4a4002dad8d5bf31_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/
  type:
    field_input: input
    field_instruction: instruct
    field_output: output
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
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: 16
gradient_checkpointing: true
group_by_length: false
hub_model_id: eason668/a13db28a-1559-4abf-bc30-e440c80187ef
learning_rate: 5.0e-05
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
lr_scheduler: cosine
max_steps: 200
micro_batch_size: 8
mlflow_experiment_name: /tmp/4a4002dad8d5bf31_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
val_set_size: 0.1
wandb_entity: null
wandb_mode: online
wandb_name: a13db28a-1559-4abf-bc30-e440c80187ef
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: a13db28a-1559-4abf-bc30-e440c80187ef
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

a13db28a-1559-4abf-bc30-e440c80187ef

This model is a fine-tuned version of samoline/651e2cac-f73e-483e-9e71-50976a75fcaa on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6095
Memory/max Mem Active(gib): 28.44
Memory/max Mem Allocated(gib): 24.04
Memory/device Mem Reserved(gib): 33.94

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: 5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 16
total_train_batch_size: 1024
total_eval_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: 100
training_steps: 200

Training results

Training Loss	Epoch	Step	Validation Loss	Mem Active(gib)	Mem Allocated(gib)	Mem Reserved(gib)
No log	0	0	0.6165	28.44	21.52	33.94
0.6352	0.4008	25	0.6164	28.44	24.04	33.94
0.6349	0.8016	50	0.6160	28.44	24.04	33.94
0.6372	1.1924	75	0.6153	28.44	24.04	33.94
0.6344	1.5932	100	0.6140	28.44	24.04	33.94
0.614	1.9940	125	0.6122	28.44	24.04	33.94
0.6124	2.3848	150	0.6107	28.44	24.04	33.94
0.6065	2.7856	175	0.6096	28.44	24.04	33.94
0.6161	3.1764	200	0.6095	28.44	24.04	33.94

Framework versions

PEFT 0.17.0
Transformers 4.55.0
Pytorch 2.6.0+cu124
Datasets 4.0.0
Tokenizers 0.21.4