Spaces:
Runtime error
Runtime error
Commit History
-
Fix AI summary format - use proper chat template
jesseliu0913- Fixed inference to properly parse SYSTEM/USER prompts - Used tokenizer.apply_chat_template() for correct formatting - Summaries now show only model output (no prompt echo) - Generated proper PREOP SUMMARY format with tables - All 24 patients have clean, properly formatted summaries81ae992 -
Fix patient IDs and add complete AI summaries
jesseliu0913- Fixed subject_id field (was null, now correct integer IDs) - All 24 patients now have complete AI-generated summaries - Generated using Llama-3.2-3B-Instruct (temp=0.2) - Fixes Clinician_2 and other panels showing 0 patients64309c9 -
Add AI-generated summaries for 24 patient samples
jesseliu0913- Generated using Llama-3.2-3B-Instruct - Temperature: 0.2, Max tokens: 1024 - All 24 samples now have pre-operative anesthesia summaries919f41a -
Update data: 24 samples across 8 length groups (500-10K chars)
jesseliu0913- Replace 20 samples with 24 systematically selected samples - Stratify across 8 length groups (group_1-3, group_4-10) - Add comprehensive documentation (DATA_UPDATE_README.md) - Add deployment guide (DEPLOY_WITH_NEW_DATA.md) - Backup original data (structured_results_o3_with_inputs.jsonl.backup) - Source: /playpen/jesse/ehr_mimic/new_benchmark_0121/2fc9f33
-
update
Jesse Liub3e7b51
-
patient notes summary demo
706667c
