EngineX/xc-llm-kunlun
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

update base image

Browse Source
This commit is contained in:
starkwj
2026-03-02 18:46:04 +08:00
parent a15754c3ba
commit 34e04c5569
3 changed files with 6 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Dockerfile
View File

@@ -1,4 +1,4 @@
FROM wjie520/vllm_kunlun:base_v0.0.2 FROM vllm_kunlun:custom_base_v0.0.3
WORKDIR /workspace WORKDIR /workspace

9
README.md
View File

@@ -11,11 +11,12 @@ One of the key features of this project is efficient memory coordination, enabli
### Build from Dockerfile ### Build from Dockerfile
Clone this repository: 1. Get or build base image (base with customized xpytorch, ops, etc.). Ref: [installation](https://vllm-kunlun.readthedocs.io/en/latest/installation.html).
```bash 2. Clone this repository and build
docker build -t $build_image -f ./Dockerfile . ```bash
``` docker build -t $build_image -f ./Dockerfile .
```
## Usage ## Usage

3
vllm_kunlun/platforms/kunlun.py
View File

@@ -8,9 +8,6 @@ import vllm.envs as envs
from vllm.logger import init_logger from vllm.logger import init_logger
# fix bfloat16 double size issue
torch.backends.cuda.matmul.allow_bf16_reduced_precision_reduction = False
logger = init_logger(__name__) logger = init_logger(__name__)
class KunlunPlatform(Platform): class KunlunPlatform(Platform):