Skip to content

Conversation

@a120092009
Copy link
Contributor

No description provided.

int32_t dp_local_tp_size_;
int32_t num_experts_per_tok_;
int32_t num_speculative_tokens_ = 0;
at::Device device_;
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

use torch::

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have removed those useless variables due to my unwary copying from xllm/models/llm/deepseek_v2.h

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fixed

#include <gflags/gflags.h>
#include <torch/torch.h>

#include <boost/algorithm/string.hpp>
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remove this line

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

removed


#include <torch/torch.h>

#include <functional>
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

it seems useless.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

removed

/*num_experts=*/model_args.n_routed_experts(),
/*top_k=*/model_args.num_experts_per_tok(),
/*num_expert_group=*/model_args.n_group(),
/*topk_group=*/model_args.topk_group(),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

no need to add such comments for these var.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

all clean now!

bool is_gated,
bool has_bias,
const std::string& hidden_act,
bool if_reduce_results,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: use bool enable_result_reduction.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

fixed

==============================================================================*/
#pragma once

#include <gflags/gflags.h>
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

seems useless

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

removed

@XuZhang99 XuZhang99 changed the title feat: build deepseek v2 decoder layer and deepseek related model files. feat: build deepseek v2 decoder layer and deepseek related model files for mlu device. Nov 13, 2025
@XuZhang99 XuZhang99 changed the title feat: build deepseek v2 decoder layer and deepseek related model files for mlu device. feat: build deepseek v2 decoder layer and related model files for mlu device. Nov 13, 2025
@a120092009 a120092009 force-pushed the mlu/feat_deepseek_layer branch from d83692b to b2f779c Compare November 14, 2025 03:57
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants