accuracy enhancement for baichuan2-13B for npu (#16868)

Co-authored-by: cy <chenyang08056032@163.com>
2026-01-26 16:14:35 +08:00
parent 6756cf18c6
commit 2734b23481
5 changed files with 241 additions and 75 deletions
@@ -464,6 +464,9 @@ class ModelConfig:
                    ):
                        setattr(self.hf_text_config, "head_dim", self.head_dim)

+            elif "BaichuanForCausalLM" in self.hf_config.architectures:
+                self.use_alibi = self.hf_config.hidden_size != 4096
+
            self.attention_arch = AttentionArch.MHA

        self.num_attention_heads = self.hf_text_config.num_attention_heads