Search results

hidden size 2048 of Qwen3-30B-A3B is not supported

Status: Open.

#233 In deepseek-ai/DeepEP;

· GuoxiangZu opened on Jun 20, 2025

Any plan of releases or stable tags?

Status: Open.

#231 In deepseek-ai/DeepEP;

· kuozhang opened on Jun 20, 2025

[Feature Request]: Support File Generation (e.g., PDF, TXT) from Model Output

Status: Open.

#230 In deepseek-ai/DeepEP;

· isaacmahyoub opened on Jun 19, 2025

Issue 1: Missing Imports in setup.py

Status: Open.

#229 In deepseek-ai/DeepEP;

· PrinceSajjadHussain opened on Jun 19, 2025

Issue 1: Missing <code>__init__.py</code> files in packages

Status: Open.

#228 In deepseek-ai/DeepEP;

· PrinceSajjadHussain opened on Jun 19, 2025

Issue 1: Missing docstring for internode_combine method

Status: Open.

#227 In deepseek-ai/DeepEP;

· PrinceSajjadHussain opened on Jun 19, 2025

By default, in megatron, condition fais on ibgda_get_state()-&gt;num_rc_per_pe == num_channels || ibgda_get_state()-&gt;num_rc_per_pe &gt;= num_sms

Status: Open.

#226 In deepseek-ai/DeepEP;

· kuozhang opened on Jun 19, 2025

What circumstances should low_latency_mode be set to false

Status: Open.

#221 In deepseek-ai/DeepEP;

· pc-neo opened on Jun 18, 2025

crash when set NVSHMEM_IBGDA_ENABLE_MULTI_PORT to 1 in case of dual-port RNIC

Status: Open.

#199 In deepseek-ai/DeepEP;

· yanminjia opened on Jun 10, 2025

When deploying with multiple machines, the following error was encountered

Status: Open.

#185 In deepseek-ai/DeepEP;

· nannaer opened on Jun 4, 2025

Can we reduce the kernel latency for normal dispatch and combine when overlapping with gemm kernels through tuning the UNROLL_FACTOR?

Status: Open.

#184 In deepseek-ai/DeepEP;

· yaning223 opened on Jun 4, 2025

Does DeepEP use SMs for its normal and low-latency kernels?

Status: Open.

#179 In deepseek-ai/DeepEP;

· MaoZiming opened on Jun 2, 2025