radlab
/

polish-qa-v2-bnb

Question Answering

8-bit precision

Model card Files Files and versions

pkedzia commited on Oct 4

Commit

0670ee9

·

verified ·

1 Parent(s): a0221cf

Update README.md

Proper base model

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -5,7 +5,7 @@ datasets:
 language:
 - pl
 base_model:
-- sdadas/polish-roberta-large-v2
 pipeline_tag: question-answering
 library_name: transformers
 tags:
@@ -20,7 +20,7 @@ tags:
 - **Model name**: `radlab/polish-qa-v2-bnb`
 - **Developer**: [radlab.dev](https://radlab.dev)
 - **Model type**: Extractive Question‑Answering (QA)
-- **Base model**: `sdadas/polish-roberta-large-v2` (fine‑tuned for QA)
 - **Quantization**: 8‑bit inference‑only quantization via **bitsandbytes** (`load_in_8bit=True`, double‑quantization enabled, `qa_outputs` excluded from quantization)
 - **Maximum context size**: 512 tokens

 language:
 - pl
 base_model:
+- radlab/polish-qa-v2
 pipeline_tag: question-answering
 library_name: transformers
 tags:
 - **Model name**: `radlab/polish-qa-v2-bnb`
 - **Developer**: [radlab.dev](https://radlab.dev)
 - **Model type**: Extractive Question‑Answering (QA)
+- **Base model**: `radlab/polish-qa-v` (`sdadas/polish-roberta-large-v2` fine‑tuned for QA)
 - **Quantization**: 8‑bit inference‑only quantization via **bitsandbytes** (`load_in_8bit=True`, double‑quantization enabled, `qa_outputs` excluded from quantization)
 - **Maximum context size**: 512 tokens