feat: support Qwen3-VL model on npu device. #295
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
wly-115
force-pushed
the
qwen3-vl
branch
5 times, most recently
from
39f3227 to
d08f719
Compare
yq33victor
reviewed
Nov 3, 2025
| const ParallelArgs& parallel_args) { | ||
| param.hasSharedExpert = (args.n_shared_experts() > 0); | ||
| param.hasSharedExpertGate = true; | ||
| param.hasSharedExpertGate = false; |
Collaborator
There was a problem hiding this comment.
other qwen models will be effected by this modification.
| const int local_index = expert_index % num_experts_per_partition_; | ||
| const bool is_sharded = shard_map.count(index); | ||
|
|
||
| std::lock_guard<std::mutex> lock(experts_mutex_); |
|
|
||
| void load_model(std::unique_ptr<ModelLoader> loader, | ||
| std::string prefix = "" /*llm model weight prefix*/) { | ||
| std::string prefix = "model." /*llm model weight prefix*/) { |
Collaborator
There was a problem hiding this comment.
nit: all models has the prefix model. ?
It seems that the llm_head does not have this prefix.
lm_head_->load_state_dict(
state_dict->get_dict_with_prefix(prefix + "lm_head."));
| return embed_tokens_(input_ids, 0); | ||
| #elif defined(USE_MLU) | ||
| return embed_tokens_(input_ids); | ||
| #endif |
Collaborator
There was a problem hiding this comment.
nit: Defensive programming is needed here.
#else
LOG(FATAL) << "...."
liutongxuan
changed the title
liutongxuan
approved these changes
Nov 4, 2025
yq33victor
approved these changes
Nov 4, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
feat:
1.support Qwen3-VL.
2.Qwen3-VL-MOE is currently not supported.