From 0fe9a916e88e884d6ee7fa20c4b15f5a9610218f Mon Sep 17 00:00:00 2001 From: ylsun <1654366371@qq.com> Date: Sat, 8 Aug 2026 19:27:13 +0800 Subject: [PATCH] [Bugfix] Fix LFM2 ShortConv prefix breaking quant ignore list (#51495) Signed-off-by: ylsun <1654366371@qq.com> Co-authored-by: Claude Opus 5 (1M context) --- vllm/model_executor/models/lfm2.py | 2 +- vllm/model_executor/models/lfm2_moe.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/vllm/model_executor/models/lfm2.py b/vllm/model_executor/models/lfm2.py index e18edb0b36..69a0e23be4 100644 --- a/vllm/model_executor/models/lfm2.py +++ b/vllm/model_executor/models/lfm2.py @@ -258,7 +258,7 @@ class Lfm2ShortConvDecoderLayer(nn.Module): model_config=model_config, cache_config=cache_config, quant_config=quant_config, - prefix=f"{prefix}.conv", + prefix=f"{prefix}.short_conv", ) self.feed_forward = Lfm2MLP( diff --git a/vllm/model_executor/models/lfm2_moe.py b/vllm/model_executor/models/lfm2_moe.py index 0b41cb0f0f..44bf55917e 100644 --- a/vllm/model_executor/models/lfm2_moe.py +++ b/vllm/model_executor/models/lfm2_moe.py @@ -354,7 +354,7 @@ class Lfm2MoeShortConvDecoderLayer(nn.Module): model_config=model_config, cache_config=cache_config, quant_config=quant_config, - prefix=f"{prefix}.conv", + prefix=f"{prefix}.short_conv", ) if layer_idx < config.num_dense_layers: