Skip to content

[quantization] Add eval_tasks option #522

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dayo09 merged 1 commit into from
Feb 26, 2026
Merged

[quantization] Add eval_tasks option #522

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions tico/quantization/wrapq/examples/quantize_full_qmodel_with_gptq.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,7 @@
import torch
import tqdm
from datasets import load_dataset
from lm_eval.utils import make_table
from transformers import AutoModelForCausalLM, AutoTokenizer

from transformers.cache_utils import Cache
Expand All @@ -48,6 +49,7 @@
from tico.quantization import convert, prepare
from tico.quantization.config.gptq import GPTQConfig
from tico.quantization.config.ptq import PTQConfig
from tico.quantization.evaluation.script.llm_tasks_eval import evaluate_llm_on_tasks
from tico.quantization.wrapq.dtypes import DType
from tico.quantization.wrapq.observers.affine_base import AffineObserverBase
from tico.quantization.wrapq.qscheme import QScheme
Expand Down Expand Up @@ -364,6 +366,11 @@ def evaluate(q_m, tokenizer, dataset_test, args):
print(f"│ int16 : {ppl_uint8:8.2f}")
print("└───────────────────────────────────────────")

if args.eval_tasks is not None:
results = evaluate_llm_on_tasks(q_m, tokenizer, args.eval_tasks)
print("Quantized RESULTS ARE:")
print(make_table(results))


def main():
parser = argparse.ArgumentParser(
Expand Down Expand Up @@ -521,6 +528,11 @@ def main():
print(f"│ FP32 : {ppl_fp32:8.2f}")
print("└───────────────────────────────────────────")

if args.eval_tasks is not None:
results = evaluate_llm_on_tasks(model, tokenizer, args.eval_tasks)
print("Original RESULTS ARE:")
print(make_table(results))

# -------------------------------------------------------------------------
# Prepare calibration dataset
# -------------------------------------------------------------------------
Expand Down