Downloads · 30 days
29
4% of all-time downloads
cstorm125/wangchan-deberta_v1-base-wiki-20210520-news-spm-finetune-qa
wangchan-deberta_v1-base-wiki-20210520-news-spm-finetune-qa is a question answering model from cstorm125. Use it when the input is a question plus a passage. It is set up for transformers.
Finetuning airesearch/wangchan-debertav1-base-wiki-20210520-news-spm with the training set of iappwikiqasquad, thaiqasquad, and nscqa (removed examples which have cosine similarity with validation and test examples ov…
Downloads · 30 days
29
4% of all-time downloads
All-time downloads
655
Public
Repo size
955 MB
Likes
0
Public
Click a slice to open those files.
.bin477 MB · 99%
From the Hugging Face model README
Finetuning airesearch/wangchan-deberta_v1-base-wiki-20210520-news-spm with the training set of iapp_wiki_qa_squad, thaiqa_squad, and nsc_qa (removed examples which have cosine similarity with validation and test examples over 0.8; contexts of the latter two are trimmed to be around 300 newmm words). Benchmarks shared on wandb using validation and test sets of iapp_wiki_qa_squad.
Trained with thai2transformers.
Run with:
export MODEL_NAME=wangchan-deberta_v1-base-wiki-20210520-news-spm
CUDA_LAUNCH_BLOCKING=1 python train_question_answering_lm_finetuning.py \
--model_name $MODEL_NAME \
--dataset_name chimera_qa \
--revision mlm@ckp-41100 \
--output_dir $MODEL_NAME-finetune-chimera_qa-model \
--log_dir $MODEL_NAME-finetune-chimera_qa-log \
--model_max_length 400 \
--pad_on_right \
--fp16 \
--use_auth_token