Spaces:
Sleeping
Sleeping
Update Dockerfile
Browse files- Dockerfile +1 -32
Dockerfile
CHANGED
|
@@ -35,35 +35,4 @@ CMD ["/app/llamafile", \
|
|
| 35 |
"--port", "7860", \
|
| 36 |
"-ngl", "0", \
|
| 37 |
"-t", "2", \
|
| 38 |
-
"-c", "4096"]
|
| 39 |
-
```
|
| 40 |
-
|
| 41 |
-
你还需要确认 `README.md` 顶部是这样：
|
| 42 |
-
|
| 43 |
-
```yaml
|
| 44 |
-
---
|
| 45 |
-
title: LLM Chat
|
| 46 |
-
emoji: 🤖
|
| 47 |
-
colorFrom: blue
|
| 48 |
-
colorTo: purple
|
| 49 |
-
sdk: docker
|
| 50 |
-
app_port: 7860
|
| 51 |
-
pinned: false
|
| 52 |
-
---
|
| 53 |
-
```
|
| 54 |
-
|
| 55 |
-
提交后等它重新 Build，成功后直接打开：
|
| 56 |
-
|
| 57 |
-
```text
|
| 58 |
-
https://huggingface.co/spaces/javaeeduke/llm
|
| 59 |
-
```
|
| 60 |
-
|
| 61 |
-
就应该能看到 llamafile 的聊天界面。
|
| 62 |
-
|
| 63 |
-
如果这个 1.5B Q8 模型启动慢，可以把模型换成更小的 0.5B：
|
| 64 |
-
|
| 65 |
-
```dockerfile
|
| 66 |
-
RUN curl -L --fail \
|
| 67 |
-
https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF/resolve/main/qwen2.5-0.5b-instruct-q8_0.gguf \
|
| 68 |
-
-o /app/model.gguf
|
| 69 |
-
```
|
|
|
|
| 35 |
"--port", "7860", \
|
| 36 |
"-ngl", "0", \
|
| 37 |
"-t", "2", \
|
| 38 |
+
"-c", "4096"]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|