Text Generation
PEFT
Safetensors
Korean
Chinese
joseon
hanmun
hanja
korean-translation
lora
qwen
historical-translation
conversational
Instructions to use suhjae/joseon-level4-qwen36-27b-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use suhjae/joseon-level4-qwen36-27b-lora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3.6-27B") model = PeftModel.from_pretrained(base_model, "suhjae/joseon-level4-qwen36-27b-lora") - Notebooks
- Google Colab
- Kaggle
Download dev_sample8_eval.json from suhjae/joseon-level4-qwen36-27b-lora: direct link, hf CLI and curl.
- Browser
- Download file 924 Bytes
-
https://huggingface.co/suhjae/joseon-level4-qwen36-27b-lora/resolve/main/dev_sample8_eval.json
- Command line
-
hf download hf://suhjae/joseon-level4-qwen36-27b-lora/dev_sample8_eval.json
-
curl -L -o dev_sample8_eval.json https://huggingface.co/suhjae/joseon-level4-qwen36-27b-lora/resolve/main/dev_sample8_eval.json
924 Bytes
| { | |
| "reference_metrics": { | |
| "chrf": null, | |
| "comet": null, | |
| "bleurt": null | |
| }, | |
| "source_grounded": { | |
| "person_reading_accuracy": 0.0, | |
| "office_accuracy": 1.0, | |
| "entity_f1": 0.0, | |
| "institutional_term_accuracy": 1.0, | |
| "rough_event_type_accuracy": 0.0, | |
| "action_accuracy": 1.0, | |
| "argument_role_f1": 0.0, | |
| "polarity_accuracy": 1.0, | |
| "hallucinated_entity_rate": 0.0, | |
| "omitted_entity_rate": 0.0, | |
| "office_reading_accuracy": 1.0, | |
| "place_reading_accuracy": 0.0, | |
| "book_reading_accuracy": 0.0, | |
| "all_reading_accuracy": 1.0, | |
| "entity_precision": 0.0, | |
| "entity_recall": 0.0, | |
| "office_title_accuracy": 1.0 | |
| }, | |
| "style": { | |
| "hanja_output_rate": 0.0, | |
| "untranslated_span_rate": 0.0, | |
| "length_ratio": 0.9556129729266671, | |
| "korean_only_output_rate": 1.0 | |
| }, | |
| "composite": { | |
| "historical_integrity": 0.7333333333333333, | |
| "overall": 0.7333333333333333 | |
| } | |
| } |