Spaces:
Sleeping
Sleeping
Update requirements.txt
Browse files- requirements.txt +18 -31
requirements.txt
CHANGED
|
@@ -1,39 +1,30 @@
|
|
| 1 |
-
#
|
| 2 |
-
|
| 3 |
-
|
| 4 |
transformers==4.40.1
|
|
|
|
|
|
|
| 5 |
peft==0.10.0
|
|
|
|
| 6 |
datasets
|
| 7 |
-
trl
|
| 8 |
-
accelerate
|
| 9 |
-
sentence-transformers
|
| 10 |
-
torch
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
# Core ML and AI
|
| 15 |
-
torch>=2.0.0
|
| 16 |
-
transformers==4.39.3
|
| 17 |
-
sentence-transformers>=2.2.2
|
| 18 |
-
accelerate>=0.24.0
|
| 19 |
-
bitsandbytes==0.41.1
|
| 20 |
|
|
|
|
|
|
|
| 21 |
|
|
|
|
| 22 |
unstructured
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
tesseract
|
| 26 |
pytesseract
|
| 27 |
pdfminer.six
|
|
|
|
| 28 |
|
| 29 |
-
|
| 30 |
-
# Vector Database and Search
|
| 31 |
faiss-cpu>=1.7.4
|
| 32 |
llama-index>=0.9.0
|
| 33 |
llama-index-vector-stores-faiss>=0.1.0
|
| 34 |
llama-index-embeddings-huggingface>=0.1.0
|
| 35 |
|
| 36 |
-
# Document
|
| 37 |
PyPDF2>=3.0.1
|
| 38 |
python-docx>=0.8.11
|
| 39 |
openpyxl>=3.1.2
|
|
@@ -43,28 +34,24 @@ gradio>=4.0.0
|
|
| 43 |
fastapi>=0.104.0
|
| 44 |
uvicorn>=0.24.0
|
| 45 |
|
| 46 |
-
# Data
|
| 47 |
numpy>=1.24.0
|
| 48 |
pandas>=2.0.0
|
| 49 |
scipy>=1.11.0
|
| 50 |
-
|
| 51 |
-
# Utilities
|
| 52 |
requests>=2.31.0
|
| 53 |
tqdm>=4.66.0
|
| 54 |
python-dotenv>=1.0.0
|
| 55 |
pydantic>=2.4.0
|
| 56 |
|
| 57 |
-
# Async
|
| 58 |
asyncio-throttle>=1.0.2
|
| 59 |
aiofiles>=23.2.1
|
| 60 |
|
| 61 |
-
# Logging
|
| 62 |
loguru>=0.7.2
|
| 63 |
|
| 64 |
-
# Optional:
|
| 65 |
# onnxruntime>=1.16.0
|
| 66 |
# optimum>=1.14.0
|
| 67 |
-
|
| 68 |
-
# Optional: For advanced quantization (uncomment if needed)
|
| 69 |
# auto-gptq>=0.4.2
|
| 70 |
-
# autoawq>=0.1.6
|
|
|
|
| 1 |
+
# Core ML and AI
|
| 2 |
+
torch>=2.0.0
|
|
|
|
| 3 |
transformers==4.40.1
|
| 4 |
+
sentence-transformers==2.5.1
|
| 5 |
+
accelerate==0.24.1
|
| 6 |
peft==0.10.0
|
| 7 |
+
trl==0.7.10
|
| 8 |
datasets
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
+
# Optional: 8-bit quantization (uncomment if needed)
|
| 11 |
+
# bitsandbytes==0.41.1
|
| 12 |
|
| 13 |
+
# Document Parsing & OCR
|
| 14 |
unstructured
|
| 15 |
+
# Optional: add specific extras like [pdf] or [local-inference] depending on your platform
|
| 16 |
+
# unstructured[pdf]
|
|
|
|
| 17 |
pytesseract
|
| 18 |
pdfminer.six
|
| 19 |
+
tesseract # Note: system dependency; install via apt, not pip
|
| 20 |
|
| 21 |
+
# Vector DB & Search
|
|
|
|
| 22 |
faiss-cpu>=1.7.4
|
| 23 |
llama-index>=0.9.0
|
| 24 |
llama-index-vector-stores-faiss>=0.1.0
|
| 25 |
llama-index-embeddings-huggingface>=0.1.0
|
| 26 |
|
| 27 |
+
# Document Formats
|
| 28 |
PyPDF2>=3.0.1
|
| 29 |
python-docx>=0.8.11
|
| 30 |
openpyxl>=3.1.2
|
|
|
|
| 34 |
fastapi>=0.104.0
|
| 35 |
uvicorn>=0.24.0
|
| 36 |
|
| 37 |
+
# Data & Utilities
|
| 38 |
numpy>=1.24.0
|
| 39 |
pandas>=2.0.0
|
| 40 |
scipy>=1.11.0
|
|
|
|
|
|
|
| 41 |
requests>=2.31.0
|
| 42 |
tqdm>=4.66.0
|
| 43 |
python-dotenv>=1.0.0
|
| 44 |
pydantic>=2.4.0
|
| 45 |
|
| 46 |
+
# Async & Performance
|
| 47 |
asyncio-throttle>=1.0.2
|
| 48 |
aiofiles>=23.2.1
|
| 49 |
|
| 50 |
+
# Logging
|
| 51 |
loguru>=0.7.2
|
| 52 |
|
| 53 |
+
# Optional: ONNX & quantization (uncomment if needed)
|
| 54 |
# onnxruntime>=1.16.0
|
| 55 |
# optimum>=1.14.0
|
|
|
|
|
|
|
| 56 |
# auto-gptq>=0.4.2
|
| 57 |
+
# autoawq>=0.1.6
|