refactor: multimodal data (#4754)

2025-03-31 09:57:51 -07:00
parent c7457191a0
commit 5cb552b1d4
36 changed files with 989 additions and 1138 deletions
@@ -1308,6 +1308,9 @@ class DeepseekV2ForCausalLM(nn.Module):
        self.logits_processor = LogitsProcessor(config)
        self.dp_size = get_attention_dp_size()

+    def get_input_embeddings(self) -> nn.Embedding:
+        return self.model.embed_tokens
+
    @torch.no_grad()
    def forward(
        self,