Kresna Sucandra
SHA888
·
AI & ML interests
None yet
Organizations
fix(train): use eval_strategy for TrainingArguments on Space
#12 opened 3 months ago
by
SHA888
feat(ui): add 'Trust dataset script' checkbox and plumb trust_remote_code for conll2003
#11 opened 3 months ago
by
SHA888
chore(defaults): set BioBERT v1.1 + conll2003 and epochs=3 as baseline
#10 opened 3 months ago
by
SHA888
Add NER LoRA checkpoint (checkpoints/ner-wikiann-bert-smoke)
#5 opened 3 months ago
by
SHA888
feat(loader): prefer BigBio script loader; fallback to parquet; clearer errors
#9 opened 3 months ago
by
SHA888
build: pin datasets<3.0 to support community dataset loaders (BigBio)
#8 opened 3 months ago
by
SHA888
fix(dataset): list parquet via HfApi and load via HTTPS data_files; add dataset dropdown
#7 opened 3 months ago
by
SHA888
feat(dataset): use Parquet hf:// for BigBio (bc5cdr, ncbi_disease) and add dataset dropdown
#6 opened 3 months ago
by
SHA888
fix(dataset): enable trust_remote_code for BigBio; robust load_dataset retry
#5 opened 3 months ago
by
SHA888
feat(train): add bc5cdr/ncbi_disease aliases; robust dataset loader and retry
#4 opened 3 months ago
by
SHA888
Consolidate: resolve app.py conflicts; finalize training Space
#3 opened 3 months ago
by
SHA888
Enhance: generic token/tag columns, metrics in PR description, publish med-vllm-* variant
#2 opened 3 months ago
by
SHA888
Create Training Space: LoRA NER trainer (Gradio UI)
#1 opened 3 months ago
by
SHA888
Fix: remove medvllm import; load config via huggingface_hub
#4 opened 3 months ago
by
SHA888
Fix build: remove non-installable medvllm; add huggingface_hub + pyyaml
#3 opened 3 months ago
by
SHA888
Fix: add Spaces YAML front matter
#2 opened 3 months ago
by
SHA888
Create Space: Med vLLM config-first demo (Gradio)
#1 opened 3 months ago
by
SHA888
Docs: enhance README with Quickstart, examples, weights roadmap
#4 opened 3 months ago
by
SHA888
Add root config.yaml (NER default)
#3 opened 3 months ago
by
SHA888
Initial config export (README, examples)
#1 opened 3 months ago
by
SHA888