Search results

Any plan to support MTP (Multi Token Prediction) based speculative decode?

Status: Open.

#32 In hemingkx/Spec-Bench;

· Stonesjtu opened on Oct 23, 2025

Support loading of <code>.bin</code>, <code>.pt</code>, <code>.pth</code> models

Status: Open.

#31 In hemingkx/Spec-Bench;

· NamburiSrinath opened on Jun 10, 2025

Is num_choices and block size same as gamma for the Speculative decoding?

Status: Open.

#30 In hemingkx/Spec-Bench;

· NamburiSrinath opened on May 21, 2025

Possible issue with speed computation - Higher #Mean Accepted tokens but lower speedups.

Status: Open.

#29 In hemingkx/Spec-Bench;

· NamburiSrinath opened on May 21, 2025

Request for Qwen2/3 series support

Status: Open.

#27 In hemingkx/Spec-Bench;

· FWXT opened on May 4, 2025

How to use multiple GPUs？

Status: Open.

#26 In hemingkx/Spec-Bench;

· Ss-scee opened on Apr 18, 2025

Why do I get ERROR question ID on questions 288 and 317 when evaluating Medusa and Hydra on specbench?

Status: Open.

#25 In hemingkx/Spec-Bench;

· Tomor1 opened on Apr 12, 2025