rrw-x2/KoSOLAR-10.7B-qlora-v1.2.1

See axolotl config

axolotl version: 0.4.0

base_model: JY623/KoSOLAR-v2.0
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: true
strict: false

push_dataset_to_hub:
datasets:
  - path: kyujinpy/KOR-OpenOrca-Platypus-v3
    type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./qlora-out/v1.2

adapter: qlora
lora_model_dir:

sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true

lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_32bit
lr_scheduler: cosine
learning_rate: 0.0002

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: false

warmup_steps: 20
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
  unk_token: "<unk>"

qlora-out/v1.2

This model is a fine-tuned version of JY623/KoSOLAR-v2.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 5.1419

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 7
• gradient_accumulation_steps: 4
• total_train_batch_size: 28
• total_eval_batch_size: 7
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 20
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
13.4775	0.0	1	13.4330
6.9219	0.25	64	6.2022
5.5416	0.5	128	5.2780
5.4282	0.75	192	5.1929
5.4864	1.0	256	5.1416
5.2877	1.24	320	5.1441
5.1731	1.49	384	5.1413
5.6221	1.74	448	5.1406
5.3737	1.99	512	5.1419

Framework versions

• PEFT 0.9.0
• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.0