RYDE-WORK/lnp_ml
1
0
Fork 0
mirror of https://github.com/RYDE-WORK/lnp_ml.git synced 2026-03-21 09:36:32 +08:00
Code Issues Packages Projects Releases Wiki Activity

...

Browse Source
This commit is contained in:
RYDE-WORK 2026-01-22 00:24:13 +08:00
parent e123fc8f3e
commit e6a5e5495a
21 changed files with 108 additions and 212 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
Makefile
View File

@ -74,9 +74,9 @@ data_pretrain: requirements
$(PYTHON_INTERPRETER) scripts/process_external.py $(PYTHON_INTERPRETER) scripts/process_external.py
## Process CV data for cross-validation pretrain (external/all_amine_split_for_LiON -> processed/cv) ## Process CV data for cross-validation pretrain (external/all_amine_split_for_LiON -> processed/cv)
.PHONY: data_cv .PHONY: data_pretrain_cv
data_cv: requirements data_pretrain_cv: requirements
$(PYTHON_INTERPRETER) scripts/process_data_cv.py $(PYTHON_INTERPRETER) scripts/process_external_cv.py
# MPNN 支持：使用 USE_MPNN=1 启用 MPNN encoder # MPNN 支持：使用 USE_MPNN=1 启用 MPNN encoder
# 例如：make pretrain USE_MPNN=1 # 例如：make pretrain USE_MPNN=1
@ -106,8 +106,8 @@ pretrain_cv: requirements
$(PYTHON_INTERPRETER) -m lnp_ml.modeling.pretrain_cv main $(MPNN_FLAG) $(DEVICE_FLAG) $(PYTHON_INTERPRETER) -m lnp_ml.modeling.pretrain_cv main $(MPNN_FLAG) $(DEVICE_FLAG)
## Evaluate CV pretrain models on test sets (auto-detects MPNN from checkpoint) ## Evaluate CV pretrain models on test sets (auto-detects MPNN from checkpoint)
.PHONY: test_cv .PHONY: test_pretrain_cv
test_cv: requirements test_pretrain_cv: requirements
$(PYTHON_INTERPRETER) -m lnp_ml.modeling.pretrain_cv test $(DEVICE_FLAG) $(PYTHON_INTERPRETER) -m lnp_ml.modeling.pretrain_cv test $(DEVICE_FLAG)
## Train model (multi-task, from scratch) ## Train model (multi-task, from scratch)

0
data/processed/cv/feature_columns.txt → data/processed/pretrain_cv/feature_columns.txt
View File

0
data/processed/cv/fold_0/test.parquet → data/processed/pretrain_cv/fold_0/test.parquet
View File

0
data/processed/cv/fold_0/train.parquet → data/processed/pretrain_cv/fold_0/train.parquet
View File

0
data/processed/cv/fold_0/valid.parquet → data/processed/pretrain_cv/fold_0/valid.parquet
View File

0
data/processed/cv/fold_1/test.parquet → data/processed/pretrain_cv/fold_1/test.parquet
View File

0
data/processed/cv/fold_1/train.parquet → data/processed/pretrain_cv/fold_1/train.parquet
View File

0
data/processed/cv/fold_1/valid.parquet → data/processed/pretrain_cv/fold_1/valid.parquet
View File

0
data/processed/cv/fold_2/test.parquet → data/processed/pretrain_cv/fold_2/test.parquet
View File

0
data/processed/cv/fold_2/train.parquet → data/processed/pretrain_cv/fold_2/train.parquet
View File

0
data/processed/cv/fold_2/valid.parquet → data/processed/pretrain_cv/fold_2/valid.parquet
View File

0
data/processed/cv/fold_3/test.parquet → data/processed/pretrain_cv/fold_3/test.parquet
View File

0
data/processed/cv/fold_3/train.parquet → data/processed/pretrain_cv/fold_3/train.parquet
View File

0
data/processed/cv/fold_3/valid.parquet → data/processed/pretrain_cv/fold_3/valid.parquet
View File

0
data/processed/cv/fold_4/test.parquet → data/processed/pretrain_cv/fold_4/test.parquet
View File

0
data/processed/cv/fold_4/train.parquet → data/processed/pretrain_cv/fold_4/train.parquet
View File

0
data/processed/cv/fold_4/valid.parquet → data/processed/pretrain_cv/fold_4/valid.parquet
View File

6
lnp_ml/modeling/pretrain_cv.py
View File

@ -271,7 +271,7 @@ def create_model(
@app.command() @app.command()
def main( def main(
data_dir: Path = PROCESSED_DATA_DIR / "cv", data_dir: Path = PROCESSED_DATA_DIR / "pretrain_cv",
output_dir: Path = MODELS_DIR / "pretrain_cv", output_dir: Path = MODELS_DIR / "pretrain_cv",
# 模型参数 # 模型参数
d_model: int = 256, d_model: int = 256,
@ -322,7 +322,7 @@ def main(
if not fold_dirs: if not fold_dirs:
logger.error(f"No fold_* directories found in {data_dir}") logger.error(f"No fold_* directories found in {data_dir}")
logger.info("Please run 'make data_cv' first to process CV data.") logger.info("Please run 'make data_pretrain_cv' first to process CV data.")
raise typer.Exit(1) raise typer.Exit(1)
logger.info(f"Found {len(fold_dirs)} folds: {[d.name for d in fold_dirs]}") logger.info(f"Found {len(fold_dirs)} folds: {[d.name for d in fold_dirs]}")
@ -464,7 +464,7 @@ def main(
@app.command() @app.command()
def test( def test(
data_dir: Path = PROCESSED_DATA_DIR / "cv", data_dir: Path = PROCESSED_DATA_DIR / "pretrain_cv",
model_dir: Path = MODELS_DIR / "pretrain_cv", model_dir: Path = MODELS_DIR / "pretrain_cv",
output_path: Path = MODELS_DIR / "pretrain_cv" / "test_results.json", output_path: Path = MODELS_DIR / "pretrain_cv" / "test_results.json",
batch_size: int = 64, batch_size: int = 64,

BIN
models/pretrain_delivery.pt
View File

Binary file not shown.

304
models/pretrain_history.json
View File

@ -1,310 +1,206 @@
{ {
"train": [ "train": [
{ {
"loss": 0.8244676398801744, "loss": 0.7730368412685099,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.6991508170533461, "loss": 0.658895703010919,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.6388374940987616, "loss": 0.6059015260392299,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.6008581508669937, "loss": 0.5744731174349416,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.584832567446085, "loss": 0.5452056020458733,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.5481657371815157, "loss": 0.5138543470936083,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.5368926340308079, "loss": 0.4885380559178135,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.5210388793613561, "loss": 0.47587182296687974,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.49758357966374045, "loss": 0.4671051038255316,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.49256294099457043, "loss": 0.46794115915756107,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4697267088016886, "loss": 0.4293930456997915,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.45763822707571084, "loss": 0.42624105651716415,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4495221330627172, "loss": 0.4131358770446828,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.446159594079631, "loss": 0.3946074267790835,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4327090857889029, "loss": 0.3898155013755344,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4249273364101852, "loss": 0.37861797005733383,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4216959138704459, "loss": 0.3775682858392304,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.416526201182502, "loss": 0.3800349080262064,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.40368679039741573, "loss": 0.36302345173031675,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.4051084730032182, "loss": 0.3429561740842766,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.38971701020385785, "loss": 0.3445638883004898,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.39155546386038786, "loss": 0.318970229203733,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.37976963541784114, "loss": 0.30179278279904437,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.36484339719805037, "loss": 0.2887343142006437,
"n_samples": 8721 "n_samples": 6783
}, },
{ {
"loss": 0.36232607571196496, "loss": 0.29240367556855545,
"n_samples": 8721 "n_samples": 6783
},
{
"loss": 0.3345973272380199,
"n_samples": 8721
},
{
"loss": 0.31767916518768957,
"n_samples": 8721
},
{
"loss": 0.32065429246052457,
"n_samples": 8721
},
{
"loss": 0.3171297926146043,
"n_samples": 8721
},
{
"loss": 0.3122120894173009,
"n_samples": 8721
},
{
"loss": 0.3135035038404461,
"n_samples": 8721
},
{
"loss": 0.2987745178222875,
"n_samples": 8721
},
{
"loss": 0.2914867957853393,
"n_samples": 8721
},
{
"loss": 0.2983839795507705,
"n_samples": 8721
},
{
"loss": 0.2826709597875678,
"n_samples": 8721
},
{
"loss": 0.2731766632569382,
"n_samples": 8721
},
{
"loss": 0.27726896305742266,
"n_samples": 8721
},
{
"loss": 0.27864557847067956,
"n_samples": 8721
} }
], ],
"val": [ "val": [
{ {
"loss": 0.7601077516012517, "loss": 0.7350345371841441,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.7119935319611901, "loss": 0.7165568811318536,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.6461842978148269, "loss": 0.7251406249862214,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.7006978391063226, "loss": 0.6836505264587159,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.6533874032943979, "loss": 0.6747132955771933,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.6413641451743611, "loss": 0.6691136244936912,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.6168395132979742, "loss": 0.6337480902323249,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.6095251602162025, "loss": 0.6600317959527934,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5887809592626905, "loss": 0.6439923948855346,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5655298325376368, "loss": 0.643800035575267,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5809201743872788, "loss": 0.6181512585221839,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5897585974912033, "loss": 0.6442458634939151,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5732012489662573, "loss": 0.6344759362359862,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5607388911786094, "loss": 0.6501405371457472,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5717580675371414, "loss": 0.6098835162990152,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5553950037657291, "loss": 0.6366627322138894,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5778171792857049, "loss": 0.6171610150646417,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5602665468127734, "loss": 0.6358801012273748,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5475307451359259, "loss": 0.6239976831059871,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.551515599314827, "loss": 0.6683828232827201,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5755438121541243, "loss": 0.6655785786478143,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5798238261811381, "loss": 0.6152775046503088,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5739961433828923, "loss": 0.6202247662153858,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5742599932540312, "loss": 0.648199727435189,
"n_samples": 969 "n_samples": 2907
}, },
{ {
"loss": 0.5834948123885382, "loss": 0.6473217075085124,
"n_samples": 969 "n_samples": 2907
},
{
"loss": 0.554078846570139,
"n_samples": 969
},
{
"loss": 0.5714933996322354,
"n_samples": 969
},
{
"loss": 0.5384107524350331,
"n_samples": 969
},
{
"loss": 0.570854394451568,
"n_samples": 969
},
{
"loss": 0.5767292551642478,
"n_samples": 969
},
{
"loss": 0.5660079547556808,
"n_samples": 969
},
{
"loss": 0.5608972411514312,
"n_samples": 969
},
{
"loss": 0.5620947442987263,
"n_samples": 969
},
{
"loss": 0.5706970894361305,
"n_samples": 969
},
{
"loss": 0.5702376298690974,
"n_samples": 969
},
{
"loss": 0.5758474825259579,
"n_samples": 969
},
{
"loss": 0.5673816067284844,
"n_samples": 969
},
{
"loss": 0.5671441179879925,
"n_samples": 969
} }
] ]
} }

0
scripts/process_data_cv.py → scripts/process_external_cv.py
View File