See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/tinyllama
bf16: auto
dataset_prepared_path: null
datasets:
- data_files:
  - 68cb8e1c19ecaf0a_train_data.json
  ds_type: json
  format: custom
  path: 68cb8e1c19ecaf0a_train_data.json
  type:
    field: null
    field_input: null
    field_instruction: prompt
    field_output: response_a
    field_system: null
    format: null
    no_input_format: null
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: null
eval_sample_packing: false
evals_per_epoch: 4
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: FatCat87/taopanda-2_1a63cef3-a8eb-43e4-9e58-ab6c9ca06368
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: true
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_bnb_8bit
output_dir: ./outputs/lora-out
pad_to_sequence_len: true
resume_from_checkpoint: null
sample_packing: true
saves_per_epoch: 1
seed: 31876
sequence_len: 4096
special_tokens: null
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
val_set_size: 0.05
wandb_entity: fatcat87-taopanda
wandb_log_model: null
wandb_mode: online
wandb_name: taopanda-2_1a63cef3-a8eb-43e4-9e58-ab6c9ca06368
wandb_project: subnet56
wandb_runid: taopanda-2_1a63cef3-a8eb-43e4-9e58-ab6c9ca06368
wandb_watch: null
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

taopanda-2_1a63cef3-a8eb-43e4-9e58-ab6c9ca06368

This model is a fine-tuned version of unsloth/tinyllama on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4257

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: 2
• eval_batch_size: 2
• seed: 31876
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
0.4045	0.0084	1	0.4087
0.3968	0.2510	30	0.4186
0.4001	0.5021	60	0.4176
0.4094	0.7531	90	0.4191
0.4024	1.0042	120	0.4255
0.4087	1.2385	150	0.4257
0.4288	1.4895	180	0.4256
0.4178	1.7406	210	0.4257
0.4069	1.9916	240	0.4257
0.4183	2.2259	270	0.4257
0.4297	2.4770	300	0.4257
0.4043	2.7280	330	0.4257
0.4144	2.9791	360	0.4257
0.4006	3.2134	390	0.4257
0.3987	3.4644	420	0.4257
0.4321	3.7155	450	0.4257

Framework versions

• PEFT 0.11.1
• Transformers 4.42.3
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1