Hugging Face's logo

Spaces:
dmartincy
/
document-translation
Runtime error

App Files Community
dmartincy commited on
Commit
2917d86
·
1 Parent(s): 9702737

Use llama.cpp

Browse files
Files changed (4) hide show
  1. Dockerfile +10 -10
  2. README.md +1 -0
  3. index.html +1 -1
  4. start-services.sh +5 -10
Dockerfile CHANGED
@@ -21,8 +21,17 @@ RUN apt-get update && apt-get install -y \
21
  clang \
22
  cuda-toolkit \
23
  nginx \
 
 
 
 
24
  && rm -rf /var/lib/apt/lists/*
25
 
 
 
 
 
 
26
  # Install Node.js and pnpm
27
  RUN curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
28
  apt-get update && \
@@ -34,12 +43,11 @@ RUN curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
34
  RUN corepack enable && corepack prepare pnpm@latest --activate
35
 
36
  # Create directories and set permissions
37
- RUN mkdir -p /tmp/llamafile && \
38
  mkdir -p $HOME/models && \
39
  mkdir -p $HOME/app && \
40
  mkdir -p $HOME/app/docauth && \
41
  mkdir -p $HOME/app/aia && \
42
- mkdir -p $HOME/.llamafile && \
43
  mkdir -p /var/cache/nginx && \
44
  mkdir -p /var/log/nginx && \
45
  mkdir -p /var/lib/nginx && \
@@ -52,14 +60,6 @@ RUN mkdir -p /tmp/llamafile && \
52
  chown -R user:user /var/run/nginx.pid && \
53
  chown -R user:user /run/nginx
54
 
55
- # Install llamafiler
56
- RUN cd /tmp/llamafile && \
57
- wget --no-check-certificate https://github.com/Mozilla-Ocho/llamafile/releases/download/0.8.17/llamafile-0.8.17.zip && \
58
- unzip llamafile-0.8.17.zip && \
59
- cp llamafile-0.8.17/bin/llamafiler /usr/local/bin/ && \
60
- chmod +x /usr/local/bin/llamafiler && \
61
- rm -rf /tmp/llamafile
62
-
63
  # Copy AI Assistant app
64
  COPY --from=pspdfkit/ai-assistant:nightly /base $HOME/app/aia
65
  COPY --chown=node:node service-config.yml /service-config.yml
 
21
  clang \
22
  cuda-toolkit \
23
  nginx \
24
+ build-essential \
25
+ cmake \
26
+ git \
27
+ libcurl4-openssl-dev \
28
  && rm -rf /var/lib/apt/lists/*
29
 
30
+ # Copy llama.cpp server files from official image
31
+ COPY --from=ghcr.io/ggerganov/llama.cpp:server-cuda /app/llama-server $HOME/app/llama-server
32
+ COPY --from=ghcr.io/ggerganov/llama.cpp:server-cuda /app/*.so* $HOME/app/
33
+ RUN chmod +x $HOME/app/llama-server
34
+
35
  # Install Node.js and pnpm
36
  RUN curl -fsSL https://deb.nodesource.com/setup_20.x | bash - && \
37
  apt-get update && \
 
43
  RUN corepack enable && corepack prepare pnpm@latest --activate
44
 
45
  # Create directories and set permissions
46
+ RUN mkdir -p /tmp/llamacpp && \
47
  mkdir -p $HOME/models && \
48
  mkdir -p $HOME/app && \
49
  mkdir -p $HOME/app/docauth && \
50
  mkdir -p $HOME/app/aia && \
 
51
  mkdir -p /var/cache/nginx && \
52
  mkdir -p /var/log/nginx && \
53
  mkdir -p /var/lib/nginx && \
 
60
  chown -R user:user /var/run/nginx.pid && \
61
  chown -R user:user /run/nginx
62
 
 
 
 
 
 
 
 
 
63
  # Copy AI Assistant app
64
  COPY --from=pspdfkit/ai-assistant:nightly /base $HOME/app/aia
65
  COPY --chown=node:node service-config.yml /service-config.yml
README.md CHANGED
@@ -6,6 +6,7 @@ colorTo: yellow
6
  sdk: docker
7
  pinned: false
8
  short_description: A WYSIWYG editor that translates text to multiple languages.
 
9
  ---
10
 
11
  Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
6
  sdk: docker
7
  pinned: false
8
  short_description: A WYSIWYG editor that translates text to multiple languages.
9
+ startup_duration_timeout: 1h
10
  ---
11
 
12
  Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
index.html CHANGED
@@ -207,7 +207,7 @@
207
  <body>
208
  <div id="loadingOverlay">
209
  <div id="loadingSpinner"></div>
210
- <div id="loadingMessage">Initializing AI models... This may take up to 10 minutes.</div>
211
  </div>
212
  <div id="app">
213
  <div style="height: 100vh; display: flex; flex-direction: column;">
 
207
  <body>
208
  <div id="loadingOverlay">
209
  <div id="loadingSpinner"></div>
210
+ <div id="loadingMessage">Initializing AI models... This may take a while.</div>
211
  </div>
212
  <div id="app">
213
  <div style="height: 100vh; display: flex; flex-direction: column;">
start-services.sh CHANGED
@@ -27,10 +27,10 @@ echo "Nginx started successfully"
27
 
28
  # Start the models
29
  echo "Starting models..."
30
- TMPDIR=/tmp/llamafiler /usr/local/bin/llamafiler --model $HOME/models/gemma-2b.gguf -ngl 999 --listen 0.0.0.0:8082 &
31
  GEMMA_PID=$!
32
 
33
- TMPDIR=/tmp/llamafiler /usr/local/bin/llamafiler --model $HOME/models/embeddings.gguf -ngl 999 --listen 0.0.0.0:8081 &
34
  EMBEDDINGS_PID=$!
35
 
36
  # Wait for models to be ready
@@ -39,15 +39,10 @@ START_TIME=$SECONDS
39
  TIMEOUT=600 # 10 minutes
40
 
41
  wait_for_models() {
42
- CHAT_TEST=$(curl -s -X POST http://127.0.0.1:8082/v1/chat/completions \
43
- -H "Content-Type: application/json" \
44
- -d '{"model":"gemma-2b","messages":[{"role":"user","content":"test"}]}')
45
 
46
- EMBED_TEST=$(curl -s -X POST http://127.0.0.1:8081/v1/embeddings \
47
- -H "Content-Type: application/json" \
48
- -d '{"input":"test"}')
49
-
50
- [[ "$CHAT_TEST" == *"content"* ]] && [[ "$EMBED_TEST" == *"embedding"* ]]
51
  }
52
 
53
  until wait_for_models; do
 
27
 
28
  # Start the models
29
  echo "Starting models..."
30
+ TMPDIR=/tmp/llamacpp ./llama-server -m $HOME/models/gemma-2b.gguf -ngl 999 --host 0.0.0.0 --port 8082 &
31
  GEMMA_PID=$!
32
 
33
+ TMPDIR=/tmp/llamacpp ./llama-server --embedding -m $HOME/models/embeddings.gguf -ngl 999 --host 0.0.0.0 --port 8081 &
34
  EMBEDDINGS_PID=$!
35
 
36
  # Wait for models to be ready
 
39
  TIMEOUT=600 # 10 minutes
40
 
41
  wait_for_models() {
42
+ CHAT_HEALTH=$(curl -s http://127.0.0.1:8082/health)
43
+ EMBED_HEALTH=$(curl -s http://127.0.0.1:8081/health)
 
44
 
45
+ [[ "$CHAT_HEALTH" == *"\"status\":\"ok\""* ]] && [[ "$EMBED_HEALTH" == *"\"status\":\"ok\""* ]]
 
 
 
 
46
  }
47
 
48
  until wait_for_models; do