adapter: qlora
base_model: TinyLlama/TinyLlama_v1.1
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 82fc59b447b3efcb_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/82fc59b447b3efcb_train_data.json
  type:
    field_instruction: question
    field_output: answer
    format: '{instruction}'
    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: 1
flash_attention: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: error577/ac3421ae-31af-46b4-bd4b-9e6cc71083de
hub_repo: null
hub_strategy: end
hub_token: null
learning_rate: 0.0001
load_in_4bit: true
load_in_8bit: false
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
max_steps: 1000
micro_batch_size: 1
mlflow_experiment_name: /tmp/82fc59b447b3efcb_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 50
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: 1
sequence_len: 1024
special_tokens:
  pad_token: </s>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.02
wandb_entity: null
wandb_mode: online
wandb_name: f62a9367-396d-43db-9468-361a83df1d0c
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: f62a9367-396d-43db-9468-361a83df1d0c
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

ac3421ae-31af-46b4-bd4b-9e6cc71083de

This model is a fine-tuned version of TinyLlama/TinyLlama_v1.1 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.4983

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: 0.0001
train_batch_size: 1
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 8
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: 1000

Training results

Training Loss	Epoch	Step	Validation Loss
1.5421	0.0010	1	1.4359
1.1304	0.0190	20	1.0104
0.6992	0.0380	40	0.7770
0.688	0.0570	60	0.7021
0.5184	0.0760	80	0.6654
0.5509	0.0950	100	0.6418
0.6378	0.1140	120	0.6281
0.6062	0.1330	140	0.6108
0.8244	0.1520	160	0.6009
0.5185	0.1710	180	0.5913
0.5699	0.1900	200	0.5827
0.4622	0.2090	220	0.5753
0.4492	0.2280	240	0.5718
0.5629	0.2470	260	0.5662
0.5465	0.2660	280	0.5632
0.376	0.2850	300	0.5571
0.4478	0.3040	320	0.5522
0.5251	0.3230	340	0.5496
0.4852	0.3420	360	0.5444
0.5344	0.3610	380	0.5419
0.5464	0.3800	400	0.5381
0.4565	0.3990	420	0.5354
0.4654	0.4181	440	0.5314
0.4963	0.4371	460	0.5277
0.5259	0.4561	480	0.5268
0.5111	0.4751	500	0.5241
0.5169	0.4941	520	0.5222
0.5947	0.5131	540	0.5183
0.5295	0.5321	560	0.5172
0.4934	0.5511	580	0.5151
0.4575	0.5701	600	0.5135
0.4981	0.5891	620	0.5115
0.4236	0.6081	640	0.5093
0.4831	0.6271	660	0.5095
0.3917	0.6461	680	0.5072
0.4254	0.6651	700	0.5056
0.4732	0.6841	720	0.5043
0.4753	0.7031	740	0.5033
0.4428	0.7221	760	0.5026
0.4353	0.7411	780	0.5011
0.4548	0.7601	800	0.5007
0.4652	0.7791	820	0.5001
0.6047	0.7981	840	0.4996
0.5564	0.8171	860	0.4993
0.4263	0.8361	880	0.4991
0.4986	0.8551	900	0.4989
0.4395	0.8741	920	0.4986
0.5258	0.8931	940	0.4984
0.7145	0.9121	960	0.4982
0.5519	0.9311	980	0.4983
0.537	0.9501	1000	0.4983

Framework versions

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