deepseekmirror/DeepSeek-VL2
1
0
Fork 0
mirror of https://github.com/deepseek-ai/DeepSeek-VL2.git synced 2025-04-19 18:19:03 -04:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge cef3ab5e28 into ef9f91e2b6

Browse Source
This commit is contained in:
Ren Yanyu 2025-02-27 17:14:31 +08:00 committed by GitHub
commit 93bb5022bb
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 5 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
README.md
View File

@ -146,6 +146,7 @@ inputs_embeds = vl_gpt.prepare_inputs_embeds(**prepare_inputs)
# run the model to get the response
outputs = vl_gpt.language.generate(
input_ids = prepare_inputs["input_ids"].to(vl_gpt.device),
inputs_embeds=inputs_embeds,
attention_mask=prepare_inputs.attention_mask,
pad_token_id=tokenizer.eos_token_id,

5
deepseek_vl2/models/modeling_deepseek.py
View File

@ -1742,13 +1742,16 @@ class DeepseekV2ForCausalLM(DeepseekV2PreTrainedModel):
output = (logits,) + outputs[1:]
return (loss,) + output if loss is not None else output
return CausalLMOutputWithPast(
device = input_ids.device if input_ids is not None else inputs_embeds.device
output = CausalLMOutputWithPast(
loss=loss,
logits=logits,
past_key_values=outputs.past_key_values,
hidden_states=outputs.hidden_states,
attentions=outputs.attentions,
)
output['logits'] = output['logits'].to(device)
return output
def prepare_inputs_for_generation(
self,