axolotl version: 0.11.0.dev0
base_model: minpeter/tiny-ko-20m-base
hub_model_id: minpeter/tiny-ko-20m-sft
output_dir: ./outputs/tiny-ko-20m-sft
wandb_project: "axolotl"
wandb_entity: "kasfiekfs-e"
chat_template: chatml
datasets:
- path: lemon-mint/Korean-FineTome-100k
type: chat_template
split: train[:1000]
field_messages: messages
message_property_mappings:
role: role
content: content
- path: lemon-mint/smol-koreantalk
type: chat_template
split: train[:1000]
field_messages: messages
message_property_mappings:
role: role
content: content
- path: heegyu/open-korean-instructions-v20231020
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
roles:
user: ["human", "user"]
assistant: ["gpt", "assistant", "bot"]
system: ["system", "input"]
# NOTE: https://github.com/FreedomIntelligence/MultilingualSIFT
- path: FreedomIntelligence/evol-instruct-korean
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
- path: FreedomIntelligence/alpaca-gpt4-korean
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
- path: FreedomIntelligence/sharegpt-korean
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
- path: coastral/korean-writing-style-instruct
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
- path: devngho/korean-instruction-mix
type: chat_template
split: train[:1000]
field_messages: messages
message_property_mappings:
role: from
content: value
- path: youjunhyeok/Magpie-Pro-300K-Filtered-ko
type: chat_template
split: train[:1000]
field_messages: conversations
message_property_mappings:
role: from
content: value
- path: youjunhyeok/smoltalk-ko-translate
type: chat_template
split: train[:1000]
name: merge_filtered
field_messages: conversations
message_property_mappings:
role: role
content: content
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
save_steps: 200
warmup_steps: 20
eval_steps: 200
sequence_len: 8192
# <<<< experimental settings <<<<
sample_packing: false
train_on_inputs: false
# >>>> experimental settings >>>
pad_to_sequence_len: true
gradient_accumulation_steps: 4
micro_batch_size: 16
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 1e-3
bf16: auto
tf32: false
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
resume_from_checkpoint:
logging_steps: 1
flash_attention: true
num_epochs: 1
weight_decay: 0.0
tiny-ko-20m-sft
This model is a fine-tuned version of minpeter/tiny-ko-20m-base on the lemon-mint/Korean-FineTome-100k, the lemon-mint/smol-koreantalk, the heegyu/open-korean-instructions-v20231020, the FreedomIntelligence/evol-instruct-korean, the FreedomIntelligence/alpaca-gpt4-korean, the FreedomIntelligence/sharegpt-korean, the coastral/korean-writing-style-instruct, the devngho/korean-instruction-mix, the youjunhyeok/Magpie-Pro-300K-Filtered-ko and the youjunhyeok/smoltalk-ko-translate datasets.
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: 0.001
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT 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: 20
- training_steps: 75
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| No log | 0 | 0 | 3.5333 |
Framework versions
- Transformers 4.52.4
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.1