optimize imports

holunda-io · Apr 25, 2024 · 7396795 · 7396795
1 parent 9cc5f06
commit 7396795
Show file tree

Hide file tree

Showing 11 changed files with 118 additions and 119 deletions.
diff --git a/bpm_ai_inference/llm/llama_cpp/llama_chat.py b/bpm_ai_inference/llm/llama_cpp/llama_chat.py
@@ -1,5 +1,6 @@
 import json
 import logging
+import os
 import re
 from typing import Dict, Any, Optional, List
 
@@ -13,6 +14,7 @@
 from bpm_ai_inference.llm.llama_cpp._constants import DEFAULT_MODEL, DEFAULT_TEMPERATURE, DEFAULT_MAX_RETRIES, \
     DEFAULT_QUANT_BALANCED
 from bpm_ai_inference.llm.llama_cpp.util import messages_to_llama_dicts
+from bpm_ai_inference.util import FORCE_OFFLINE_FLAG
 from bpm_ai_inference.util.files import find_file
 from bpm_ai_inference.util.hf import hf_home
 
@@ -40,7 +42,7 @@ def __init__(
         filename: str = DEFAULT_QUANT_BALANCED,
         temperature: float = DEFAULT_TEMPERATURE,
         max_retries: int = DEFAULT_MAX_RETRIES,
-        force_offline: bool = False
+        force_offline: bool = os.getenv(FORCE_OFFLINE_FLAG, False)
     ):
         if not has_llama_cpp_python:
             raise ImportError('llama-cpp-python is not installed')
@@ -52,8 +54,9 @@ def __init__(
         )
         n_ctx = 4096
         if force_offline:
+            model_file = find_file(hf_home() + "hub/models--" + model.replace("/", "--"), filename)
             self.llm = Llama(
-                model_path=find_file(hf_home() + "hub/models--" + model.replace("/", "--"), filename),
+                model_path=model_file,
                 n_ctx=n_ctx,
                 verbose=False
             )
@@ -65,7 +68,6 @@ def __init__(
                 verbose=False
             )
 
-
     async def _generate_message(
         self,
         messages: List[ChatMessage],

diff --git a/bpm_ai_inference/ocr/tesseract.py b/bpm_ai_inference/ocr/tesseract.py
@@ -1,4 +1,3 @@
-import asyncio
 import logging
 import os
 import urllib

diff --git a/bpm_ai_inference/pos/spacy_pos_tagger.py b/bpm_ai_inference/pos/spacy_pos_tagger.py
@@ -1,4 +1,3 @@
-import asyncio
 import logging
 
 from bpm_ai_core.pos.pos_tagger import POSTagger, POSResult

diff --git a/bpm_ai_inference/question_answering/pix2struct_vqa.py b/bpm_ai_inference/question_answering/pix2struct_vqa.py
@@ -1,4 +1,3 @@
-import asyncio
 import logging
 
 from bpm_ai_core.llm.common.blob import Blob

diff --git a/bpm_ai_inference/question_answering/transformers_docvqa.py b/bpm_ai_inference/question_answering/transformers_docvqa.py
@@ -1,4 +1,3 @@
-import asyncio
 import logging
 
 from bpm_ai_core.llm.common.blob import Blob

diff --git a/bpm_ai_inference/speech_recognition/faster_whisper.py b/bpm_ai_inference/speech_recognition/faster_whisper.py
@@ -1,4 +1,3 @@
-import asyncio
 import io
 
 from bpm_ai_core.speech_recognition.asr import ASRModel, ASRResult

diff --git a/bpm_ai_inference/translation/easy_nmt/easy_nmt.py b/bpm_ai_inference/translation/easy_nmt/easy_nmt.py
@@ -1,4 +1,3 @@
-import asyncio
 import json
 import logging
 import math

diff --git a/bpm_ai_inference/util/__init__.py b/bpm_ai_inference/util/__init__.py
@@ -0,0 +1 @@
+FORCE_OFFLINE_FLAG = "FORCE_OFFLINE"
diff --git a/bpm_ai_inference/util/optimum.py b/bpm_ai_inference/util/optimum.py
@@ -9,7 +9,7 @@
 from huggingface_hub import HfFileSystem
 from optimum.onnxruntime import ORTModelForSequenceClassification, ORTOptimizer, ORTQuantizer, \
     ORTModelForQuestionAnswering, ORTModel
-from optimum.onnxruntime.configuration import OptimizationConfig, AutoQuantizationConfig, AutoOptimizationConfig
+from optimum.onnxruntime.configuration import AutoQuantizationConfig, AutoOptimizationConfig
 from transformers import AutoTokenizer
 
 from bpm_ai_inference.util.hf import hf_home

diff --git a/poetry.lock b/poetry.lock
diff --git a/pyproject.toml b/pyproject.toml
@@ -10,12 +10,12 @@ readme = "README.md"
 
 [tool.poetry.dependencies]
 python = ">=3.11,<3.12"
-bpm-ai-core = "^2.4.2"
+bpm-ai-core = "^2.5.0"
 langfuse = "^2.7.6"
 faster-whisper = "^0.10.0"
 lingua-language-detector = "^2.0.2"
 pytesseract = "^0.3.10"
-transformers = "^4.37.2"
+transformers = "^4.39.3"
 sacremoses = "^0.1.1"
 sentencepiece = "^0.2.0"
 nltk = "^3.8.0"