See axolotl config

axolotl version: 0.4.1

adapter: lora
auto_find_batch_size: false
base_model: aisingapore/llama3-8b-cpt-sea-lionv2.1-instruct
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - c0bbe39ceb13871b_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/c0bbe39ceb13871b_train_data.json
  type:
    field_input: full_label
    field_instruction: label
    field_output: text
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
early_stopping_threshold: 1.0e-05
eval_max_new_tokens: 128
eval_steps: 61
eval_strategy: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 3
gradient_checkpointing: false
group_by_length: false
hub_model_id: mrferr3t/8542c5fa-dea4-4467-8af5-12a941e67bde
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0004
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 61
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_steps: 
micro_batch_size: 6
mlflow_experiment_name: /tmp/c0bbe39ceb13871b_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 100
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: 61
saves_per_epoch: 0
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: .05000000
wandb_entity: null
wandb_mode: 
wandb_name: 0e0efa1d-335c-42c5-811f-ba1deee68e7e
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 0e0efa1d-335c-42c5-811f-ba1deee68e7e
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

8542c5fa-dea4-4467-8af5-12a941e67bde

This model is a fine-tuned version of aisingapore/llama3-8b-cpt-sea-lionv2.1-instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0158

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: 0.0004
• train_batch_size: 6
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 18
• 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
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0033	1	2.4959
2.2656	0.1991	61	2.1335
2.0816	0.3983	122	2.0720
2.0409	0.5974	183	2.0352
2.0112	0.7965	244	2.0075
1.9894	0.9956	305	1.9935
1.8311	1.1948	366	2.0044
1.8276	1.3939	427	1.9931
1.8365	1.5930	488	1.9885
1.8314	1.7922	549	1.9785
1.829	1.9913	610	1.9705
1.5978	2.1904	671	2.0216
1.6011	2.3896	732	2.0344
1.6257	2.5887	793	2.0158

Framework versions

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