quantize linear (#250)

2025-10-23 22:18:06 +08:00 · 2024-01-07 18:48:59 -08:00
parent 737b4c81a3
commit 485fb9ac0f
2 changed files with 0 additions and 5 deletions
--- a/lora/convert.py
+++ b/lora/convert.py
@@ -28,8 +28,6 @@ def quantize(weights, config, args):
        model,
        args.q_group_size,
        args.q_bits,
-        linear_class_predicate=lambda m: isinstance(m, nn.Linear)
-        and m.weight.shape[0] != config["vocab_size"],
    )

    # Update the config: