Update Dockerfile
Browse files- Dockerfile +10 -11
Dockerfile
CHANGED
@@ -27,22 +27,21 @@ RUN pip install --no-cache-dir \
|
|
27 |
--index-url https://download.pytorch.org/whl/cu121
|
28 |
|
29 |
EXPOSE 7860
|
30 |
-
|
31 |
-
RUN pip install --no-cache-dir text-generation==0.6.1
|
32 |
-
|
33 |
ENV HF_HOME=/app/.cache
|
|
|
|
|
|
|
|
|
34 |
ENV MODEL_ID="SebastianBodza/Kartoffel_Orpheus-3B_german_natural-v0.1"
|
35 |
ENV ORPHEUS_API_URL=http://127.0.0.1:1234/v1/completions
|
36 |
-
ENV ORPHEUS_MODEL=$MODEL_ID # landet im JSON-Payload von OrpheusEngine
|
37 |
|
|
|
38 |
CMD bash -c "\
|
39 |
-
|
40 |
-
--model
|
41 |
--port 1234 \
|
42 |
-
--
|
43 |
-
--dtype auto \
|
44 |
-
--quantization fp8 \
|
45 |
-
--enable-chunked-prefill \
|
46 |
--gpu-memory-utilization 0.7 & \
|
47 |
-
|
|
|
48 |
|
|
|
27 |
--index-url https://download.pytorch.org/whl/cu121
|
28 |
|
29 |
EXPOSE 7860
|
|
|
|
|
|
|
30 |
ENV HF_HOME=/app/.cache
|
31 |
+
# βββββ Pakete βββββββββββββββββββββββββββββββββββββββββββββ
|
32 |
+
RUN pip install --no-cache-dir vllm==0.4.2
|
33 |
+
|
34 |
+
# βββββ Environment βββββββββββββββββββββββββββββββββββββββ
|
35 |
ENV MODEL_ID="SebastianBodza/Kartoffel_Orpheus-3B_german_natural-v0.1"
|
36 |
ENV ORPHEUS_API_URL=http://127.0.0.1:1234/v1/completions
|
|
|
37 |
|
38 |
+
# βββββ Entrypoint ββββββββββββββββββββββββββββββββββββββββ
|
39 |
CMD bash -c "\
|
40 |
+
python -m vllm.entrypoints.openai.api_server \
|
41 |
+
--model $MODEL_ID \
|
42 |
--port 1234 \
|
43 |
+
--dtype bfloat16 \
|
|
|
|
|
|
|
44 |
--gpu-memory-utilization 0.7 & \
|
45 |
+
uvicorn app:app --host 0.0.0.0 --port 7860"
|
46 |
+
|
47 |
|