Skip to content

paper_pipeline.md

Latest commit

 

History

History
92 lines (68 loc) · 3.31 KB

paper_pipeline.md

File metadata and controls

92 lines (68 loc) · 3.31 KB
Raw

Paper Pipeline

This document maps paper claims to commands and configs.

Environment

conda create -n pathosynvlm python=3.11 -y
conda activate pathosynvlm
export PYTHONNOUSERSITE=1
pip install -e .

On clusters, run the commands inside an allocated compute session. Keeping PYTHONNOUSERSITE=1 avoids accidentally importing packages from a pre-existing user site.

Set local storage roots once before running the pipeline. The repo-local defaults are:

source configs/paths.example.env

For existing external storage, override only the roots that differ:

export PATHOSYNVLM_RAW_DATA_ROOT="/data/pathosynvlm/raw"
export PATHOSYNVLM_EMBEDDINGS_ROOT="/features/pathosynvlm/conch_embeddings"
export PATHOSYNVLM_STAGE1_METADATA_DIR="/outputs/pathosynvlm/stage1_metadata"
export PATHOSYNVLM_HISTAI_METADATA_DIR="/outputs/pathosynvlm/histai_metadata"
export PATHOSYNVLM_RUNS_ROOT="/outputs/pathosynvlm/runs"

Stage 1 Baseline

Config: configs/stage1_alignment_paper.json

Path fields in the config JSON use $PATHOSYNVLM_* notation. The scripts do not read these JSON files automatically; expand or replace those values if your launcher reads JSON directly instead of using the shell commands below.

python scripts/train_stage1_alignment.py \
  --metadata_json "$PATHOSYNVLM_STAGE1_METADATA_DIR/merged_metadata_3datasets_filtered_conch_v15.json" \
  --dataset_embeddings_root "$PATHOSYNVLM_EMBEDDINGS_ROOT" \
  --datasets histgen,reg_dataset \
  --patch_level 5x_512 \
  --output_dir "$PATHOSYNVLM_RUNS_ROOT/stage1_alignment"

Reported Stage 1 baseline metrics:

ROUGE-L METEOR BLEU-4 BERTScore F1
0.4743 0.4810 0.1247 0.4253

Stage 2 Main Result

Config: configs/stage2_main_paper.json

python scripts/train_stage2_histai.py \
  --metadata_standardized_json "$PATHOSYNVLM_HISTAI_METADATA_DIR/standardized_metadata_fixed_filtered_5x_512.json" \
  --dataset_embeddings_root "$PATHOSYNVLM_EMBEDDINGS_ROOT" \
  --aligner_init "$PATHOSYNVLM_RUNS_ROOT/stage1_alignment/best_aligner_weights.pt" \
  --output_dir "$PATHOSYNVLM_RUNS_ROOT/stage2_main" \
  --prompt_style double \
  --max_text_length 384 \
  --max_vision_tokens 4096 \
  --use_wsi_markers \
  --unfreeze_llm_base \
  --gradient_checkpoint

Evaluate:

python scripts/evaluate_checkpoint.py \
  --finetune_run_dir "$PATHOSYNVLM_RUNS_ROOT/stage2_main" \
  --dataset_scope histai \
  --histai_metadata_standardized_json "$PATHOSYNVLM_HISTAI_METADATA_DIR/standardized_metadata_fixed_filtered_5x_512.json" \
  --dataset_embeddings_root "$PATHOSYNVLM_EMBEDDINGS_ROOT" \
  --output_json "$PATHOSYNVLM_RUNS_ROOT/stage2_main/eval_histai.json"

Reported Stage 2 main metrics:

ROUGE-L METEOR BLEU-4 BERTScore F1 Diagnosis Exact Diagnosis Relaxed Certainty
0.2495 0.1988 0.0525 0.3018 0.1667 0.3333 0.9000

The training code logs sacreBLEU as a percentage, so 5.2512 in JSON corresponds to 0.0525 in the paper.

Ablations

Paper ablation values are stored in configs/reported_results.json. The corresponding command settings are summarized in configs/stage2_wsi_marker_ablation.json.