gfjiogopdfgdfs
commited on
Commit
•
592d153
1
Parent(s):
e8787df
Update Dockerfile
Browse files- Dockerfile +19 -5
Dockerfile
CHANGED
@@ -1,5 +1,19 @@
|
|
1 |
-
FROM nvidia/cuda:
|
2 |
-
RUN apt update && apt install git python3-pip -y
|
3 |
-
|
4 |
-
|
5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
FROM nvidia/cuda:11.8.0-devel-ubuntu22.04
|
2 |
+
RUN apt update && apt install git build-essential libopenblas-dev wget python3-pip -y
|
3 |
+
RUN apt update && apt install -y wget
|
4 |
+
ARG MODEL_PART_A
|
5 |
+
ARG MODEL_PART_B
|
6 |
+
ARG MODEL_NAME
|
7 |
+
ARG ADDITIONAL
|
8 |
+
RUN mkdir /opt/koboldcpp
|
9 |
+
RUN apt update && apt install git build-essential libopenblas-dev wget python3-pip -y
|
10 |
+
RUN git clone https://github.com/lostruins/koboldcpp /opt/koboldcpp
|
11 |
+
WORKDIR /opt/koboldcpp
|
12 |
+
COPY default.json /opt/koboldcpp/default.json
|
13 |
+
RUN make LLAMA_OPENBLAS=1 LLAMA_CUBLAS=1 LLAMA_PORTABLE=1
|
14 |
+
RUN wget -O part_a.gguf $MODEL_PART_A
|
15 |
+
RUN wget -O part_b.gguf $MODEL_PART_B
|
16 |
+
RUN cat part_a.gguf part_b.gguf > model.gguf
|
17 |
+
RUN rm part_a.gguf part_b.gguf
|
18 |
+
CMD /bin/python3 ./koboldcpp.py --model model.gguf $ADDITIONAL --port 7860 --hordeconfig $MODEL_NAME 1 1 --preloadstory default.json
|
19 |
+
|