Spaces:
Running
Running
Patryk Ptasiński
commited on
Commit
·
0b5e578
1
Parent(s):
3d96b43
Fix nvidia/NV-Embed-v2 to use trust_remote_code=True
Browse files- CLAUDE.md +5 -1
- test_models.sh +45 -0
CLAUDE.md
CHANGED
|
@@ -71,4 +71,8 @@ result = client.predict("text to embed", "model-name", api_name="/predict")
|
|
| 71 |
- Deployed on Hugging Face Spaces at https://huggingface.co/spaces/ipepe/nomic-embeddings
|
| 72 |
- Runs on port 7860
|
| 73 |
- Uses Gradio 4.36.1 (newer versions available)
|
| 74 |
-
- PyTorch configured for CPU-only via `--extra-index-url` in requirements.txt
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
- Deployed on Hugging Face Spaces at https://huggingface.co/spaces/ipepe/nomic-embeddings
|
| 72 |
- Runs on port 7860
|
| 73 |
- Uses Gradio 4.36.1 (newer versions available)
|
| 74 |
+
- PyTorch configured for CPU-only via `--extra-index-url` in requirements.txt
|
| 75 |
+
|
| 76 |
+
## Development Constraints
|
| 77 |
+
|
| 78 |
+
- There is no python installed locally, everything needs to be deployed to hugging face first
|
test_models.sh
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
|
| 3 |
+
# Test script for all embedding models
|
| 4 |
+
BASE_URL="https://ipepe-nomic-embeddings.hf.space"
|
| 5 |
+
TEST_TEXT="Hello world test"
|
| 6 |
+
|
| 7 |
+
echo "Testing all embedding models..."
|
| 8 |
+
echo "================================="
|
| 9 |
+
|
| 10 |
+
# Get list of models
|
| 11 |
+
MODELS=$(curl -s "${BASE_URL}/models" | grep -o '"[^"]*"' | grep -E "(nomic|BAAI|sentence|Snowflake|granite|Qwen|stella|nvidia|Alibaba|intfloat)" | tr -d '"')
|
| 12 |
+
|
| 13 |
+
# Test each model
|
| 14 |
+
for model in $MODELS; do
|
| 15 |
+
echo "Testing: $model"
|
| 16 |
+
|
| 17 |
+
# Test with 30 second timeout
|
| 18 |
+
response=$(timeout 30 curl -X POST "${BASE_URL}/embed" \
|
| 19 |
+
-H "Content-Type: application/json" \
|
| 20 |
+
-d "{\"text\": \"$TEST_TEXT\", \"model\": \"$model\"}" \
|
| 21 |
+
-w "\nHTTP_STATUS:%{http_code}" \
|
| 22 |
+
-s 2>/dev/null)
|
| 23 |
+
|
| 24 |
+
if [ $? -eq 124 ]; then
|
| 25 |
+
echo " ❌ TIMEOUT (>30s)"
|
| 26 |
+
else
|
| 27 |
+
status=$(echo "$response" | grep "HTTP_STATUS" | cut -d: -f2)
|
| 28 |
+
|
| 29 |
+
if [ "$status" = "200" ]; then
|
| 30 |
+
# Check if response contains embedding
|
| 31 |
+
if echo "$response" | grep -q '"embedding":\['; then
|
| 32 |
+
echo " ✅ SUCCESS"
|
| 33 |
+
else
|
| 34 |
+
echo " ⚠️ PARTIAL - No embedding in response"
|
| 35 |
+
fi
|
| 36 |
+
else
|
| 37 |
+
# Extract error message
|
| 38 |
+
error_msg=$(echo "$response" | grep -o '"error":"[^"]*"' | cut -d'"' -f4)
|
| 39 |
+
echo " ❌ ERROR ($status): $error_msg"
|
| 40 |
+
fi
|
| 41 |
+
fi
|
| 42 |
+
echo ""
|
| 43 |
+
done
|
| 44 |
+
|
| 45 |
+
echo "Testing complete!"
|