From a3bd8aba7d7a766c7a1ac6b8fe6a4afab37dc9fd Mon Sep 17 00:00:00 2001 From: Ronald Tse Date: Mon, 7 Sep 2026 18:01:29 +0200 Subject: [PATCH] fix(golden): every model's test pairs on their true volume --- src/gpu/modal_golden_matrix.py | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/src/gpu/modal_golden_matrix.py b/src/gpu/modal_golden_matrix.py index 898b734..7df7f97 100644 --- a/src/gpu/modal_golden_matrix.py +++ b/src/gpu/modal_golden_matrix.py @@ -31,17 +31,20 @@ MODELS_VOLUME = modal.Volume.from_name("secryst-models") DATASETS = modal.Volume.from_name("rababa-datasets") SECRYST_DATASETS = modal.Volume.from_name("secryst-datasets") +URDU_G2P = modal.Volume.from_name("urdu-g2p-datasets") +URDU_DIAC = modal.Volume.from_name("urdu-diacrit-datasets") +PERSIAN = modal.Volume.from_name("persian-g2p-datasets") app = modal.App("interscript-ml-golden-matrix", image=IMAGE) # index id -> volume zip path + test pairs (volume-relative) SOURCES = { "khm-latn-1.0": ("imf/khm-latn/khm-latn-1.0-fp32.zip", "/secryst-datasets/khmer-translit/test.jsonl"), - "urd-g2p-1.0": ("imf/urd-g2p/urd-g2p-1.0-fp32.zip", "urdu-g2p/test.jsonl"), - "urd-diac-1.0": ("imf/urd-diac/urd-diac-1.0-fp32.zip", "urdu-diacrit/test.jsonl"), - "tha-g2p-base-1.0": ("imf/tha-g2p-base/tha-g2p-base-1.0-fp32.zip", "thai-ipa/test.jsonl"), - "tha-g2p-small-1.0": ("imf/tha-g2p-small/tha-g2p-small-1.0-int8.zip", "thai-ipa/test.jsonl"), - "fas-g2p-1.0": ("imf/fas-g2p/fas-g2p-1.0-fp32.zip", None), + "urd-g2p-1.0": ("imf/urd-g2p/urd-g2p-1.0-fp32.zip", "/ud-g2p/urdu-g2p/test.jsonl"), + "urd-diac-1.0": ("imf/urd-diac/urd-diac-1.0-fp32.zip", "/ud-diacrit/urdu-diacrit/test.jsonl"), + "tha-g2p-base-1.0": ("imf/tha-g2p-base/tha-g2p-base-1.0-fp32.zip", "/secryst-datasets/thai-ipa/test.jsonl"), + "tha-g2p-small-1.0": ("imf/tha-g2p-small/tha-g2p-small-1.0-int8.zip", "/secryst-datasets/thai-ipa/test.jsonl"), + "fas-g2p-1.0": ("imf/fas-g2p/fas-g2p-1.0-fp32.zip", "/persian/persian_g2p/test.jsonl"), "heb-diac-1.0": ("imf/heb-diac/heb-diac-1.0-fp32.zip", "nakdimon/test-imf.jsonl"), "heb-diac-1.1": ("imf/heb-diac/heb.zip", "nakdimon/test-imf.jsonl"), "heb-diac-small-1.0": ("imf/heb-diac-small/heb-diac-small-1.0-fp32.zip", "nakdimon/test-imf.jsonl"), @@ -53,7 +56,14 @@ @app.function(cpu=8, memory=24 * 1024, timeout=4 * 3600, - volumes={"/outputs": MODELS_VOLUME, "/datasets": DATASETS, "/secryst-datasets": SECRYST_DATASETS}) + volumes={ + "/outputs": MODELS_VOLUME, + "/datasets": DATASETS, + "/secryst-datasets": SECRYST_DATASETS, + "/ud-g2p": URDU_G2P, + "/ud-diacrit": URDU_DIAC, + "/persian": PERSIAN, + }) def generate(n: int = 25) -> dict: import hashlib import json