[PR #898] [MERGED] Release 0.2.3post2 #1510

New issue

Closed

opened 2026-10-29 14:53:03 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-29 14:53:03 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/kvcache-ai/ktransformers/pull/898
Author: @KMSorSMS
Created: 3/15/2025
Status: ✅ Merged
Merged: 3/15/2025
Merged by: @KMSorSMS

Base: main ← Head: develop-0.2.3post2

📝 Commits (10+)

21fca5a Add compat layer from llama.cpp
4cda454 Don't add CUDA version to version in case not for CUDA
ae76a72 gptq_marlin: temporarily disable on AMD ROCm
38e5dbc Fix symbol lookup
c1f13a6 Correctly import compat layer from llama.cpp
c009512 Merge branch 'main' into hip
086a9d1 Add vendor control
ed84374 merge main; Add torch q8 linear
c38e77d Merge branch 'hip' of https://github.com/fxzjshm/ktransformers into hip
e5b001d Update readme; Format code; Add example yaml.

📊 Changes

32 files changed (+1717 additions, -115 deletions)

View changed files

📝 README.md (+1 -0)
📝 doc/README.md (+1 -0)
📝 doc/SUMMARY.md (+1 -0)
➕ doc/en/ROCm.md (+96 -0)
📝 ktransformers/__init__.py (+1 -1)
📝 ktransformers/ktransformers_ext/CMakeLists.txt (+39 -0)
📝 ktransformers/ktransformers_ext/cpu_backend/cpuinfer.h (+80 -76)
📝 ktransformers/ktransformers_ext/cpu_backend/vendors/cuda.h (+13 -1)
➕ ktransformers/ktransformers_ext/cpu_backend/vendors/hip.h (+172 -0)
📝 ktransformers/ktransformers_ext/cpu_backend/vendors/musa.h (+131 -3)
➕ ktransformers/ktransformers_ext/cpu_backend/vendors/vendor.h (+13 -0)
📝 ktransformers/ktransformers_ext/cuda/custom_gguf/dequant.cu (+4 -0)
📝 ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin.cu (+1 -1)
📝 ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin.cuh (+1 -1)
📝 ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin_dtypes.cuh (+5 -0)
📝 ktransformers/ktransformers_ext/ext_bindings.cpp (+0 -1)
➕ ktransformers/ktransformers_ext/vendors/cuda.h (+15 -0)
➕ ktransformers/ktransformers_ext/vendors/hip.h (+172 -0)
➕ ktransformers/ktransformers_ext/vendors/musa.h (+137 -0)
➕ ktransformers/ktransformers_ext/vendors/vendor.h (+13 -0)

...and 12 more files

📄 Description

No description provided

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/kvcache-ai/ktransformers/pull/898 **Author:** [@KMSorSMS](https://github.com/KMSorSMS) **Created:** 3/15/2025 **Status:** ✅ Merged **Merged:** 3/15/2025 **Merged by:** [@KMSorSMS](https://github.com/KMSorSMS) **Base:** `main` ← **Head:** `develop-0.2.3post2` --- ### 📝 Commits (10+) - [`21fca5a`](https://github.com/kvcache-ai/ktransformers/commit/21fca5a326097de6629098ede47357b899868010) Add compat layer from llama.cpp - [`4cda454`](https://github.com/kvcache-ai/ktransformers/commit/4cda45433f5d464fcddba0eebb69175acf71eee1) Don't add CUDA version to version in case not for CUDA - [`ae76a72`](https://github.com/kvcache-ai/ktransformers/commit/ae76a729d872ece70952fb63c8fe497fa6787863) gptq_marlin: temporarily disable on AMD ROCm - [`38e5dbc`](https://github.com/kvcache-ai/ktransformers/commit/38e5dbc8955116e899a16f632b319468c970d05e) Fix symbol lookup - [`c1f13a6`](https://github.com/kvcache-ai/ktransformers/commit/c1f13a69ed535c0c5e0b8960c3066fd4ca4badb8) Correctly import compat layer from llama.cpp - [`c009512`](https://github.com/kvcache-ai/ktransformers/commit/c009512a93704aaa02db2877b65cc8e661b2824c) Merge branch 'main' into hip - [`086a9d1`](https://github.com/kvcache-ai/ktransformers/commit/086a9d1cceff60f709d6236262e44aa7b6b6aae8) Add vendor control - [`ed84374`](https://github.com/kvcache-ai/ktransformers/commit/ed8437413b535cbdf552d726323b8cab3a22f2b1) merge main; Add torch q8 linear - [`c38e77d`](https://github.com/kvcache-ai/ktransformers/commit/c38e77de6bad57a55378efcd141cb6aabac5eee7) Merge branch 'hip' of https://github.com/fxzjshm/ktransformers into hip - [`e5b001d`](https://github.com/kvcache-ai/ktransformers/commit/e5b001d76fba6da67987ca0c0d6e699226a633a1) Update readme; Format code; Add example yaml. ### 📊 Changes **32 files changed** (+1717 additions, -115 deletions) <details> <summary>View changed files</summary> 📝 `README.md` (+1 -0) 📝 `doc/README.md` (+1 -0) 📝 `doc/SUMMARY.md` (+1 -0) ➕ `doc/en/ROCm.md` (+96 -0) 📝 `ktransformers/__init__.py` (+1 -1) 📝 `ktransformers/ktransformers_ext/CMakeLists.txt` (+39 -0) 📝 `ktransformers/ktransformers_ext/cpu_backend/cpuinfer.h` (+80 -76) 📝 `ktransformers/ktransformers_ext/cpu_backend/vendors/cuda.h` (+13 -1) ➕ `ktransformers/ktransformers_ext/cpu_backend/vendors/hip.h` (+172 -0) 📝 `ktransformers/ktransformers_ext/cpu_backend/vendors/musa.h` (+131 -3) ➕ `ktransformers/ktransformers_ext/cpu_backend/vendors/vendor.h` (+13 -0) 📝 `ktransformers/ktransformers_ext/cuda/custom_gguf/dequant.cu` (+4 -0) 📝 `ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin.cu` (+1 -1) 📝 `ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin.cuh` (+1 -1) 📝 `ktransformers/ktransformers_ext/cuda/gptq_marlin/gptq_marlin_dtypes.cuh` (+5 -0) 📝 `ktransformers/ktransformers_ext/ext_bindings.cpp` (+0 -1) ➕ `ktransformers/ktransformers_ext/vendors/cuda.h` (+15 -0) ➕ `ktransformers/ktransformers_ext/vendors/hip.h` (+172 -0) ➕ `ktransformers/ktransformers_ext/vendors/musa.h` (+137 -0) ➕ `ktransformers/ktransformers_ext/vendors/vendor.h` (+13 -0) _...and 12 more files_ </details> ### 📄 Description _No description provided_ --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-29 14:53:03 +00:00