adapter: lora
base_model: tokyotech-llm/Llama-3-Swallow-8B-v0.1
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 9c49a90e891b9714_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/9c49a90e891b9714_train_data.json
  type:
    field_input: sectionParentTitre
    field_instruction: title_main
    field_output: texte
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device: cuda
early_stopping_patience: 1
eval_max_new_tokens: 128
eval_steps: 5
eval_table_size: null
evals_per_epoch: null
flash_attention: false
fp16: false
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: true
hub_model_id: infogeo/2b475fde-44fd-49a6-93d5-803a74ac289e
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: true
local_rank: null
logging_steps: 3
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_memory:
  0: 79GiB
max_steps: 30
micro_batch_size: 4
mlflow_experiment_name: /tmp/9c49a90e891b9714_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 1
optim_args:
  adam_beta1: 0.9
  adam_beta2: 0.95
  adam_epsilon: 1e-5
optimizer: adamw_torch
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 10
sequence_len: 1024
special_tokens:
  pad_token: <|end_of_text|>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: true
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: 805a6eb5-7b76-4b63-9885-eafd72345255
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 805a6eb5-7b76-4b63-9885-eafd72345255
warmup_steps: 5
weight_decay: 0.001
xformers_attention: true

2b475fde-44fd-49a6-93d5-803a74ac289e

This model is a fine-tuned version of tokyotech-llm/Llama-3-Swallow-8B-v0.1 on the None dataset. It achieves the following results on the evaluation set:

Loss: 1.3599

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: 4
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=adam_beta1=0.9,adam_beta2=0.95,adam_epsilon=1e-5
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 5
training_steps: 30

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0018	1	2.0801
1.4726	0.0092	5	1.6983
1.4575	0.0183	10	1.4396
1.3948	0.0275	15	1.3988
1.3388	0.0367	20	1.3738
1.4294	0.0458	25	1.3613
1.433	0.0550	30	1.3599

Framework versions

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