openlit 1.9.0__tar.gz → 1.11.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {openlit-1.9.0 → openlit-1.11.0}/PKG-INFO +12 -11
- {openlit-1.9.0 → openlit-1.11.0}/README.md +11 -10
- {openlit-1.9.0 → openlit-1.11.0}/pyproject.toml +1 -1
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/__init__.py +6 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/chroma/chroma.py +23 -21
- openlit-1.11.0/src/openlit/instrumentation/embedchain/__init__.py +55 -0
- openlit-1.11.0/src/openlit/instrumentation/embedchain/embedchain.py +165 -0
- openlit-1.11.0/src/openlit/instrumentation/gpt4all/__init__.py +52 -0
- openlit-1.11.0/src/openlit/instrumentation/gpt4all/gpt4all.py +352 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/llamaindex/__init__.py +1 -1
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/qdrant/qdrant.py +5 -5
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/semcov/__init__.py +8 -0
- {openlit-1.9.0 → openlit-1.11.0}/LICENSE +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/__helpers.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/anthropic/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/anthropic/anthropic.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/anthropic/async_anthropic.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/bedrock/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/bedrock/bedrock.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/chroma/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/cohere/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/cohere/cohere.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/groq/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/groq/async_groq.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/groq/groq.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/haystack/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/haystack/haystack.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/langchain/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/langchain/langchain.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/llamaindex/llamaindex.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/milvus/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/milvus/milvus.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/mistral/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/mistral/async_mistral.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/mistral/mistral.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/ollama/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/ollama/async_ollama.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/ollama/ollama.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/openai/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/openai/async_azure_openai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/openai/async_openai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/openai/azure_openai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/openai/openai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/pinecone/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/pinecone/pinecone.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/qdrant/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/transformers/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/transformers/transformers.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/vertexai/__init__.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/vertexai/async_vertexai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/instrumentation/vertexai/vertexai.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/otel/metrics.py +0 -0
- {openlit-1.9.0 → openlit-1.11.0}/src/openlit/otel/tracing.py +0 -0
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.1
|
2
2
|
Name: openlit
|
3
|
-
Version: 1.
|
3
|
+
Version: 1.11.0
|
4
4
|
Summary: OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications, facilitating the integration of observability into your GenAI-driven projects
|
5
5
|
Home-page: https://github.com/openlit/openlit/tree/main/openlit/python
|
6
6
|
Keywords: OpenTelemetry,otel,otlp,llm,tracing,openai,anthropic,claude,cohere,llm monitoring,observability,monitoring,gpt,Generative AI,chatGPT
|
@@ -51,18 +51,19 @@ This project adheres to the [Semantic Conventions](https://github.com/open-telem
|
|
51
51
|
|
52
52
|
## Auto Instrumentation Capabilities
|
53
53
|
|
54
|
-
| LLMs
|
55
|
-
|
56
|
-
| [✅ OpenAI](https://docs.openlit.io/latest/integrations/openai)
|
57
|
-
| [✅ Ollama](https://docs.openlit.io/latest/integrations/ollama)
|
58
|
-
| [✅ Anthropic](https://docs.openlit.io/latest/integrations/anthropic)
|
59
|
-
| [✅
|
60
|
-
| [✅
|
61
|
-
| [✅
|
54
|
+
| LLMs | Vector DBs | Frameworks |
|
55
|
+
|-----------------------------------------------------------------|----------------------------------------------|----------------------------------------------|
|
56
|
+
| [✅ OpenAI](https://docs.openlit.io/latest/integrations/openai) | [✅ ChromaDB](https://docs.openlit.io/latest/integrations/chromadb) | [✅ Langchain](https://docs.openlit.io/latest/integrations/langchain) |
|
57
|
+
| [✅ Ollama](https://docs.openlit.io/latest/integrations/ollama) | [✅ Pinecone](https://docs.openlit.io/latest/integrations/pinecone) | [✅ LiteLLM](https://docs.openlit.io/latest/integrations/litellm) |
|
58
|
+
| [✅ Anthropic](https://docs.openlit.io/latest/integrations/anthropic) | [✅ Qdrant](https://docs.openlit.io/latest/integrations/qdrant) | [✅ LlamaIndex](https://docs.openlit.io/latest/integrations/llama-index) |
|
59
|
+
| [✅ GPT4All](https://docs.openlit.io/latest/integrations/gpt4all) | [✅ Milvus](https://docs.openlit.io/latest/integrations/milvus) | [✅ Haystack](https://docs.openlit.io/latest/integrations/haystack) |
|
60
|
+
| [✅ Cohere](https://docs.openlit.io/latest/integrations/cohere) | | [✅ EmbedChain](https://docs.openlit.io/latest/integrations/embedchain) |
|
61
|
+
| [✅ Mistral](https://docs.openlit.io/latest/integrations/mistral) | |
|
62
|
+
| [✅ Azure OpenAI](https://docs.openlit.io/latest/integrations/azure-openai) | |
|
62
63
|
| [✅ HuggingFace Transformers](https://docs.openlit.io/latest/integrations/huggingface) | |
|
63
|
-
| [✅ Amazon Bedrock](https://docs.openlit.io/latest/integrations/bedrock)
|
64
|
+
| [✅ Amazon Bedrock](https://docs.openlit.io/latest/integrations/bedrock) | |
|
64
65
|
| [✅ Vertex AI](https://docs.openlit.io/latest/integrations/vertexai) | |
|
65
|
-
| [✅ Groq](https://docs.openlit.io/latest/integrations/groq)
|
66
|
+
| [✅ Groq](https://docs.openlit.io/latest/integrations/groq) |
|
66
67
|
|
67
68
|
## Supported Destinations
|
68
69
|
- [✅ OpenTelemetry Collector](https://docs.openlit.io/latest/connections/otelcol)
|
@@ -26,18 +26,19 @@ This project adheres to the [Semantic Conventions](https://github.com/open-telem
|
|
26
26
|
|
27
27
|
## Auto Instrumentation Capabilities
|
28
28
|
|
29
|
-
| LLMs
|
30
|
-
|
31
|
-
| [✅ OpenAI](https://docs.openlit.io/latest/integrations/openai)
|
32
|
-
| [✅ Ollama](https://docs.openlit.io/latest/integrations/ollama)
|
33
|
-
| [✅ Anthropic](https://docs.openlit.io/latest/integrations/anthropic)
|
34
|
-
| [✅
|
35
|
-
| [✅
|
36
|
-
| [✅
|
29
|
+
| LLMs | Vector DBs | Frameworks |
|
30
|
+
|-----------------------------------------------------------------|----------------------------------------------|----------------------------------------------|
|
31
|
+
| [✅ OpenAI](https://docs.openlit.io/latest/integrations/openai) | [✅ ChromaDB](https://docs.openlit.io/latest/integrations/chromadb) | [✅ Langchain](https://docs.openlit.io/latest/integrations/langchain) |
|
32
|
+
| [✅ Ollama](https://docs.openlit.io/latest/integrations/ollama) | [✅ Pinecone](https://docs.openlit.io/latest/integrations/pinecone) | [✅ LiteLLM](https://docs.openlit.io/latest/integrations/litellm) |
|
33
|
+
| [✅ Anthropic](https://docs.openlit.io/latest/integrations/anthropic) | [✅ Qdrant](https://docs.openlit.io/latest/integrations/qdrant) | [✅ LlamaIndex](https://docs.openlit.io/latest/integrations/llama-index) |
|
34
|
+
| [✅ GPT4All](https://docs.openlit.io/latest/integrations/gpt4all) | [✅ Milvus](https://docs.openlit.io/latest/integrations/milvus) | [✅ Haystack](https://docs.openlit.io/latest/integrations/haystack) |
|
35
|
+
| [✅ Cohere](https://docs.openlit.io/latest/integrations/cohere) | | [✅ EmbedChain](https://docs.openlit.io/latest/integrations/embedchain) |
|
36
|
+
| [✅ Mistral](https://docs.openlit.io/latest/integrations/mistral) | |
|
37
|
+
| [✅ Azure OpenAI](https://docs.openlit.io/latest/integrations/azure-openai) | |
|
37
38
|
| [✅ HuggingFace Transformers](https://docs.openlit.io/latest/integrations/huggingface) | |
|
38
|
-
| [✅ Amazon Bedrock](https://docs.openlit.io/latest/integrations/bedrock)
|
39
|
+
| [✅ Amazon Bedrock](https://docs.openlit.io/latest/integrations/bedrock) | |
|
39
40
|
| [✅ Vertex AI](https://docs.openlit.io/latest/integrations/vertexai) | |
|
40
|
-
| [✅ Groq](https://docs.openlit.io/latest/integrations/groq)
|
41
|
+
| [✅ Groq](https://docs.openlit.io/latest/integrations/groq) |
|
41
42
|
|
42
43
|
## Supported Destinations
|
43
44
|
- [✅ OpenTelemetry Collector](https://docs.openlit.io/latest/connections/otelcol)
|
@@ -1,6 +1,6 @@
|
|
1
1
|
[tool.poetry]
|
2
2
|
name = "openlit"
|
3
|
-
version = "1.
|
3
|
+
version = "1.11.0"
|
4
4
|
description = "OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications, facilitating the integration of observability into your GenAI-driven projects"
|
5
5
|
authors = ["OpenLIT"]
|
6
6
|
repository = "https://github.com/openlit/openlit/tree/main/openlit/python"
|
@@ -21,9 +21,11 @@ from openlit.instrumentation.bedrock import BedrockInstrumentor
|
|
21
21
|
from openlit.instrumentation.vertexai import VertexAIInstrumentor
|
22
22
|
from openlit.instrumentation.groq import GroqInstrumentor
|
23
23
|
from openlit.instrumentation.ollama import OllamaInstrumentor
|
24
|
+
from openlit.instrumentation.gpt4all import GPT4AllInstrumentor
|
24
25
|
from openlit.instrumentation.langchain import LangChainInstrumentor
|
25
26
|
from openlit.instrumentation.llamaindex import LlamaIndexInstrumentor
|
26
27
|
from openlit.instrumentation.haystack import HaystackInstrumentor
|
28
|
+
from openlit.instrumentation.embedchain import EmbedChainInstrumentor
|
27
29
|
from openlit.instrumentation.chroma import ChromaInstrumentor
|
28
30
|
from openlit.instrumentation.pinecone import PineconeInstrumentor
|
29
31
|
from openlit.instrumentation.qdrant import QdrantInstrumentor
|
@@ -160,9 +162,11 @@ def init(environment="default", application_name="default", tracer=None, otlp_en
|
|
160
162
|
"vertexai": "vertexai",
|
161
163
|
"groq": "groq",
|
162
164
|
"ollama": "ollama",
|
165
|
+
"gpt4all": "gpt4all",
|
163
166
|
"langchain": "langchain",
|
164
167
|
"llama_index": "llama_index",
|
165
168
|
"haystack": "haystack",
|
169
|
+
"embedchain": "embedchain",
|
166
170
|
"chroma": "chromadb",
|
167
171
|
"pinecone": "pinecone",
|
168
172
|
"qdrant": "qdrant_client",
|
@@ -214,9 +218,11 @@ def init(environment="default", application_name="default", tracer=None, otlp_en
|
|
214
218
|
"vertexai": VertexAIInstrumentor(),
|
215
219
|
"groq": GroqInstrumentor(),
|
216
220
|
"ollama": OllamaInstrumentor(),
|
221
|
+
"gpt4all": GPT4AllInstrumentor(),
|
217
222
|
"langchain": LangChainInstrumentor(),
|
218
223
|
"llama_index": LlamaIndexInstrumentor(),
|
219
224
|
"haystack": HaystackInstrumentor(),
|
225
|
+
"embedchain": EmbedChainInstrumentor(),
|
220
226
|
"chroma": ChromaInstrumentor(),
|
221
227
|
"pinecone": PineconeInstrumentor(),
|
222
228
|
"qdrant": QdrantInstrumentor(),
|
@@ -16,11 +16,13 @@ def object_count(obj):
|
|
16
16
|
"""
|
17
17
|
Counts Length of object if it exists, Else returns None
|
18
18
|
"""
|
19
|
+
try:
|
20
|
+
cnt = len(obj)
|
21
|
+
# pylint: disable=bare-except
|
22
|
+
except:
|
23
|
+
cnt = 0
|
19
24
|
|
20
|
-
|
21
|
-
return len(obj)
|
22
|
-
|
23
|
-
return None
|
25
|
+
return cnt
|
24
26
|
|
25
27
|
def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
26
28
|
tracer, pricing_info, trace_content, metrics, disable_metrics):
|
@@ -87,24 +89,24 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
87
89
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
88
90
|
SemanticConvetion.DB_OPERATION_ADD)
|
89
91
|
span.set_attribute(SemanticConvetion.DB_ID_COUNT,
|
90
|
-
object_count(kwargs.get("ids")))
|
92
|
+
object_count(kwargs.get("ids", [])))
|
91
93
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
92
|
-
object_count(kwargs.get("embeddings")))
|
94
|
+
object_count(kwargs.get("embeddings", [])))
|
93
95
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
94
|
-
object_count(kwargs.get("metadatas")))
|
96
|
+
object_count(kwargs.get("metadatas", [])))
|
95
97
|
span.set_attribute(SemanticConvetion.DB_DOCUMENTS_COUNT,
|
96
|
-
object_count(kwargs.get("documents")))
|
98
|
+
object_count(kwargs.get("documents", [])))
|
97
99
|
|
98
100
|
elif gen_ai_endpoint == "chroma.get":
|
99
101
|
db_operation = SemanticConvetion.DB_OPERATION_GET
|
100
102
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
101
103
|
SemanticConvetion.DB_OPERATION_GET)
|
102
104
|
span.set_attribute(SemanticConvetion.DB_ID_COUNT,
|
103
|
-
object_count(kwargs.get("ids")))
|
105
|
+
object_count(kwargs.get("ids", [])))
|
104
106
|
span.set_attribute(SemanticConvetion.DB_QUERY_LIMIT,
|
105
|
-
kwargs.get("limit"))
|
107
|
+
kwargs.get("limit", ""))
|
106
108
|
span.set_attribute(SemanticConvetion.DB_OFFSET,
|
107
|
-
kwargs.get("offset"))
|
109
|
+
kwargs.get("offset", ""))
|
108
110
|
span.set_attribute(SemanticConvetion.DB_WHERE_DOCUMENT,
|
109
111
|
str(kwargs.get("where_document", "")))
|
110
112
|
|
@@ -113,7 +115,7 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
113
115
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
114
116
|
SemanticConvetion.DB_OPERATION_QUERY)
|
115
117
|
span.set_attribute(SemanticConvetion.DB_STATEMENT,
|
116
|
-
str(kwargs.get("query_texts")))
|
118
|
+
str(kwargs.get("query_texts", "")))
|
117
119
|
span.set_attribute(SemanticConvetion.DB_N_RESULTS,
|
118
120
|
kwargs.get("n_results", ""))
|
119
121
|
span.set_attribute(SemanticConvetion.DB_FILTER,
|
@@ -126,33 +128,33 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
126
128
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
127
129
|
SemanticConvetion.DB_OPERATION_UPDATE)
|
128
130
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
129
|
-
object_count(kwargs.get("embeddings")))
|
131
|
+
object_count(kwargs.get("embeddings", [])))
|
130
132
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
131
|
-
object_count(kwargs.get("metadatas")))
|
133
|
+
object_count(kwargs.get("metadatas", [])))
|
132
134
|
span.set_attribute(SemanticConvetion.DB_ID_COUNT,
|
133
|
-
object_count(kwargs.get("ids")))
|
135
|
+
object_count(kwargs.get("ids", [])))
|
134
136
|
span.set_attribute(SemanticConvetion.DB_DOCUMENTS_COUNT,
|
135
|
-
object_count(kwargs.get("documents")))
|
137
|
+
object_count(kwargs.get("documents", [])))
|
136
138
|
|
137
139
|
elif gen_ai_endpoint == "chroma.upsert":
|
138
140
|
db_operation = SemanticConvetion.DB_OPERATION_UPSERT
|
139
141
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
140
142
|
SemanticConvetion.DB_OPERATION_UPSERT)
|
141
143
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
142
|
-
object_count(kwargs.get("embeddings")))
|
144
|
+
object_count(kwargs.get("embeddings", [])))
|
143
145
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
144
|
-
object_count(kwargs.get("metadatas")))
|
146
|
+
object_count(kwargs.get("metadatas", [])))
|
145
147
|
span.set_attribute(SemanticConvetion.DB_ID_COUNT,
|
146
|
-
object_count(kwargs.get("ids")))
|
148
|
+
object_count(kwargs.get("ids", [])))
|
147
149
|
span.set_attribute(SemanticConvetion.DB_DOCUMENTS_COUNT,
|
148
|
-
object_count(kwargs.get("documents")))
|
150
|
+
object_count(kwargs.get("documents", [])))
|
149
151
|
|
150
152
|
elif gen_ai_endpoint == "chroma.delete":
|
151
153
|
db_operation = SemanticConvetion.DB_OPERATION_DELETE
|
152
154
|
span.set_attribute(SemanticConvetion.DB_OPERATION,
|
153
155
|
SemanticConvetion.DB_OPERATION_DELETE)
|
154
156
|
span.set_attribute(SemanticConvetion.DB_ID_COUNT,
|
155
|
-
object_count(kwargs.get("ids")))
|
157
|
+
object_count(kwargs.get("ids", [])))
|
156
158
|
span.set_attribute(SemanticConvetion.DB_FILTER,
|
157
159
|
str(kwargs.get("where", "")))
|
158
160
|
span.set_attribute(SemanticConvetion.DB_DELETE_ALL,
|
@@ -0,0 +1,55 @@
|
|
1
|
+
# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
|
2
|
+
"""Initializer of Auto Instrumentation of EmbedChain Functions"""
|
3
|
+
from typing import Collection
|
4
|
+
import importlib.metadata
|
5
|
+
from opentelemetry.instrumentation.instrumentor import BaseInstrumentor
|
6
|
+
from wrapt import wrap_function_wrapper
|
7
|
+
|
8
|
+
from openlit.instrumentation.embedchain.embedchain import evaluate, get_data_sources
|
9
|
+
|
10
|
+
_instruments = ("embedchain >= 0.1.104",)
|
11
|
+
|
12
|
+
WRAPPED_METHODS = [
|
13
|
+
{
|
14
|
+
"package": "embedchain",
|
15
|
+
"object": "App.evaluate",
|
16
|
+
"endpoint": "embedchain.evaluate",
|
17
|
+
"wrapper": evaluate,
|
18
|
+
},
|
19
|
+
{
|
20
|
+
"package": "embedchain",
|
21
|
+
"object": "App.get_data_sources",
|
22
|
+
"endpoint": "embedchain.get_data_sources",
|
23
|
+
"wrapper": get_data_sources,
|
24
|
+
},
|
25
|
+
]
|
26
|
+
|
27
|
+
class EmbedChainInstrumentor(BaseInstrumentor):
|
28
|
+
"""An instrumentor for EmbedChain's client library."""
|
29
|
+
|
30
|
+
def instrumentation_dependencies(self) -> Collection[str]:
|
31
|
+
return _instruments
|
32
|
+
|
33
|
+
def _instrument(self, **kwargs):
|
34
|
+
application_name = kwargs.get("application_name")
|
35
|
+
environment = kwargs.get("environment")
|
36
|
+
tracer = kwargs.get("tracer")
|
37
|
+
pricing_info = kwargs.get("pricing_info")
|
38
|
+
trace_content = kwargs.get("trace_content")
|
39
|
+
version = importlib.metadata.version("embedchain")
|
40
|
+
|
41
|
+
for wrapped_method in WRAPPED_METHODS:
|
42
|
+
wrap_package = wrapped_method.get("package")
|
43
|
+
wrap_object = wrapped_method.get("object")
|
44
|
+
gen_ai_endpoint = wrapped_method.get("endpoint")
|
45
|
+
wrapper = wrapped_method.get("wrapper")
|
46
|
+
wrap_function_wrapper(
|
47
|
+
wrap_package,
|
48
|
+
wrap_object,
|
49
|
+
wrapper(gen_ai_endpoint, version, environment, application_name,
|
50
|
+
tracer, pricing_info, trace_content),
|
51
|
+
)
|
52
|
+
|
53
|
+
@staticmethod
|
54
|
+
def _uninstrument(self, **kwargs):
|
55
|
+
pass
|
@@ -0,0 +1,165 @@
|
|
1
|
+
# pylint: disable=duplicate-code, broad-exception-caught, too-many-statements, unused-argument
|
2
|
+
"""
|
3
|
+
Module for monitoring EmbedChain applications.
|
4
|
+
"""
|
5
|
+
|
6
|
+
import logging
|
7
|
+
from opentelemetry.trace import SpanKind, Status, StatusCode
|
8
|
+
from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
|
9
|
+
from openlit.__helpers import handle_exception
|
10
|
+
from openlit.semcov import SemanticConvetion
|
11
|
+
|
12
|
+
# Initialize logger for logging potential issues and operations
|
13
|
+
logger = logging.getLogger(__name__)
|
14
|
+
|
15
|
+
def evaluate(gen_ai_endpoint, version, environment, application_name,
|
16
|
+
tracer, pricing_info, trace_content):
|
17
|
+
"""
|
18
|
+
Creates a wrapper around a function call to trace and log its execution metrics.
|
19
|
+
|
20
|
+
This function wraps any given function to measure its execution time,
|
21
|
+
log its operation, and trace its execution using OpenTelemetry.
|
22
|
+
|
23
|
+
Parameters:
|
24
|
+
- gen_ai_endpoint (str): A descriptor or name for the endpoint being traced.
|
25
|
+
- version (str): The version of the EmbedChain application.
|
26
|
+
- environment (str): The deployment environment (e.g., 'production', 'development').
|
27
|
+
- application_name (str): Name of the EmbedChain application.
|
28
|
+
- tracer (opentelemetry.trace.Tracer): The tracer object used for OpenTelemetry tracing.
|
29
|
+
- pricing_info (dict): Information about the pricing for internal metrics (currently not used).
|
30
|
+
- trace_content (bool): Flag indicating whether to trace the content of the response.
|
31
|
+
|
32
|
+
Returns:
|
33
|
+
- function: A higher-order function that takes a function 'wrapped' and returns
|
34
|
+
a new function that wraps 'wrapped' with additional tracing and logging.
|
35
|
+
"""
|
36
|
+
|
37
|
+
def wrapper(wrapped, instance, args, kwargs):
|
38
|
+
"""
|
39
|
+
An inner wrapper function that executes the wrapped function, measures execution
|
40
|
+
time, and records trace data using OpenTelemetry.
|
41
|
+
|
42
|
+
Parameters:
|
43
|
+
- wrapped (Callable): The original function that this wrapper will execute.
|
44
|
+
- instance (object): The instance to which the wrapped function belongs. This
|
45
|
+
is used for instance methods. For static and classmethods,
|
46
|
+
this may be None.
|
47
|
+
- args (tuple): Positional arguments passed to the wrapped function.
|
48
|
+
- kwargs (dict): Keyword arguments passed to the wrapped function.
|
49
|
+
|
50
|
+
Returns:
|
51
|
+
- The result of the wrapped function call.
|
52
|
+
|
53
|
+
The wrapper initiates a span with the provided tracer, sets various attributes
|
54
|
+
on the span based on the function's execution and response, and ensures
|
55
|
+
errors are handled and logged appropriately.
|
56
|
+
"""
|
57
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
58
|
+
response = wrapped(*args, **kwargs)
|
59
|
+
|
60
|
+
try:
|
61
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
62
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
63
|
+
SemanticConvetion.GEN_AI_SYSTEM_EMBEDCHAIN)
|
64
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
65
|
+
gen_ai_endpoint)
|
66
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
|
67
|
+
environment)
|
68
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
69
|
+
SemanticConvetion.GEN_AI_TYPE_FRAMEWORK)
|
70
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
71
|
+
application_name)
|
72
|
+
span.set_attribute(SemanticConvetion.GEN_AI_EVAL_CONTEXT_RELEVANCY,
|
73
|
+
response["context_relevancy"])
|
74
|
+
span.set_attribute(SemanticConvetion.GEN_AI_EVAL_ANSWER_RELEVANCY,
|
75
|
+
response["answer_relevancy"])
|
76
|
+
span.set_attribute(SemanticConvetion.GEN_AI_EVAL_GROUNDEDNESS,
|
77
|
+
response["groundedness"])
|
78
|
+
|
79
|
+
span.set_status(Status(StatusCode.OK))
|
80
|
+
|
81
|
+
# Return original response
|
82
|
+
return response
|
83
|
+
|
84
|
+
except Exception as e:
|
85
|
+
handle_exception(span, e)
|
86
|
+
logger.error("Error in trace creation: %s", e)
|
87
|
+
|
88
|
+
# Return original response
|
89
|
+
return response
|
90
|
+
|
91
|
+
return wrapper
|
92
|
+
|
93
|
+
def get_data_sources(gen_ai_endpoint, version, environment, application_name,
|
94
|
+
tracer, pricing_info, trace_content):
|
95
|
+
"""
|
96
|
+
Creates a wrapper around a function call to trace and log its execution metrics.
|
97
|
+
|
98
|
+
This function wraps any given function to measure its execution time,
|
99
|
+
log its operation, and trace its execution using OpenTelemetry.
|
100
|
+
|
101
|
+
Parameters:
|
102
|
+
- gen_ai_endpoint (str): A descriptor or name for the endpoint being traced.
|
103
|
+
- version (str): The version of the EmbedChain application.
|
104
|
+
- environment (str): The deployment environment (e.g., 'production', 'development').
|
105
|
+
- application_name (str): Name of the EmbedChain application.
|
106
|
+
- tracer (opentelemetry.trace.Tracer): The tracer object used for OpenTelemetry tracing.
|
107
|
+
- pricing_info (dict): Information about the pricing for internal metrics (currently not used).
|
108
|
+
- trace_content (bool): Flag indicating whether to trace the content of the response.
|
109
|
+
|
110
|
+
Returns:
|
111
|
+
- function: A higher-order function that takes a function 'wrapped' and returns
|
112
|
+
a new function that wraps 'wrapped' with additional tracing and logging.
|
113
|
+
"""
|
114
|
+
|
115
|
+
def wrapper(wrapped, instance, args, kwargs):
|
116
|
+
"""
|
117
|
+
An inner wrapper function that executes the wrapped function, measures execution
|
118
|
+
time, and records trace data using OpenTelemetry.
|
119
|
+
|
120
|
+
Parameters:
|
121
|
+
- wrapped (Callable): The original function that this wrapper will execute.
|
122
|
+
- instance (object): The instance to which the wrapped function belongs. This
|
123
|
+
is used for instance methods. For static and classmethods,
|
124
|
+
this may be None.
|
125
|
+
- args (tuple): Positional arguments passed to the wrapped function.
|
126
|
+
- kwargs (dict): Keyword arguments passed to the wrapped function.
|
127
|
+
|
128
|
+
Returns:
|
129
|
+
- The result of the wrapped function call.
|
130
|
+
|
131
|
+
The wrapper initiates a span with the provided tracer, sets various attributes
|
132
|
+
on the span based on the function's execution and response, and ensures
|
133
|
+
errors are handled and logged appropriately.
|
134
|
+
"""
|
135
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
136
|
+
response = wrapped(*args, **kwargs)
|
137
|
+
|
138
|
+
try:
|
139
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
140
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
141
|
+
SemanticConvetion.GEN_AI_SYSTEM_EMBEDCHAIN)
|
142
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
143
|
+
gen_ai_endpoint)
|
144
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
|
145
|
+
environment)
|
146
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
147
|
+
SemanticConvetion.GEN_AI_TYPE_FRAMEWORK)
|
148
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
149
|
+
application_name)
|
150
|
+
span.set_attribute(SemanticConvetion.GEN_AI_DATA_SOURCES,
|
151
|
+
len(response))
|
152
|
+
|
153
|
+
span.set_status(Status(StatusCode.OK))
|
154
|
+
|
155
|
+
# Return original response
|
156
|
+
return response
|
157
|
+
|
158
|
+
except Exception as e:
|
159
|
+
handle_exception(span, e)
|
160
|
+
logger.error("Error in trace creation: %s", e)
|
161
|
+
|
162
|
+
# Return original response
|
163
|
+
return response
|
164
|
+
|
165
|
+
return wrapper
|
@@ -0,0 +1,52 @@
|
|
1
|
+
# pylint: disable=useless-return, bad-staticmethod-argument, disable=duplicate-code
|
2
|
+
"""Initializer of Auto Instrumentation of GPT4All Functions"""
|
3
|
+
|
4
|
+
from typing import Collection
|
5
|
+
import importlib.metadata
|
6
|
+
from opentelemetry.instrumentation.instrumentor import BaseInstrumentor
|
7
|
+
from wrapt import wrap_function_wrapper
|
8
|
+
|
9
|
+
from openlit.instrumentation.gpt4all.gpt4all import (
|
10
|
+
embed, generate
|
11
|
+
)
|
12
|
+
|
13
|
+
_instruments = ("gpt4all >= 2.6.0",)
|
14
|
+
|
15
|
+
class GPT4AllInstrumentor(BaseInstrumentor):
|
16
|
+
"""
|
17
|
+
An instrumentor for GPT4All's client library.
|
18
|
+
"""
|
19
|
+
|
20
|
+
def instrumentation_dependencies(self) -> Collection[str]:
|
21
|
+
return _instruments
|
22
|
+
|
23
|
+
def _instrument(self, **kwargs):
|
24
|
+
application_name = kwargs.get("application_name", "default_application")
|
25
|
+
environment = kwargs.get("environment", "default_environment")
|
26
|
+
tracer = kwargs.get("tracer")
|
27
|
+
metrics = kwargs.get("metrics_dict")
|
28
|
+
pricing_info = kwargs.get("pricing_info", {})
|
29
|
+
trace_content = kwargs.get("trace_content", False)
|
30
|
+
disable_metrics = kwargs.get("disable_metrics")
|
31
|
+
version = importlib.metadata.version("gpt4all")
|
32
|
+
|
33
|
+
# generate
|
34
|
+
wrap_function_wrapper(
|
35
|
+
"gpt4all",
|
36
|
+
"GPT4All.generate",
|
37
|
+
generate("gpt4all.generate", version, environment, application_name,
|
38
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics),
|
39
|
+
)
|
40
|
+
|
41
|
+
# embed
|
42
|
+
wrap_function_wrapper(
|
43
|
+
"gpt4all",
|
44
|
+
"Embed4All.embed",
|
45
|
+
embed("gpt4all.embed", version, environment, application_name,
|
46
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics),
|
47
|
+
)
|
48
|
+
|
49
|
+
|
50
|
+
def _uninstrument(self, **kwargs):
|
51
|
+
# Proper uninstrumentation logic to revert patched methods
|
52
|
+
pass
|
@@ -0,0 +1,352 @@
|
|
1
|
+
# pylint: disable=duplicate-code, broad-exception-caught, too-many-statements, unused-argument, possibly-used-before-assignment
|
2
|
+
"""
|
3
|
+
Module for monitoring GPT4All API calls.
|
4
|
+
"""
|
5
|
+
|
6
|
+
import logging
|
7
|
+
from opentelemetry.trace import SpanKind, Status, StatusCode
|
8
|
+
from opentelemetry.sdk.resources import TELEMETRY_SDK_NAME
|
9
|
+
from openlit.__helpers import handle_exception, general_tokens
|
10
|
+
from openlit.semcov import SemanticConvetion
|
11
|
+
|
12
|
+
# Initialize logger for logging potential issues and operations
|
13
|
+
logger = logging.getLogger(__name__)
|
14
|
+
|
15
|
+
def generate(gen_ai_endpoint, version, environment, application_name,
|
16
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics):
|
17
|
+
"""
|
18
|
+
Generates a telemetry wrapper for generate to collect metrics.
|
19
|
+
|
20
|
+
Args:
|
21
|
+
gen_ai_endpoint: Endpoint identifier for logging and tracing.
|
22
|
+
version: Version of the monitoring package.
|
23
|
+
environment: Deployment environment (e.g., production, staging).
|
24
|
+
application_name: Name of the application using the GPT4All API.
|
25
|
+
tracer: OpenTelemetry tracer for creating spans.
|
26
|
+
pricing_info: Information used for calculating the cost of GPT4All usage.
|
27
|
+
trace_content: Flag indicating whether to trace the actual content.
|
28
|
+
|
29
|
+
Returns:
|
30
|
+
A function that wraps the generate method to add telemetry.
|
31
|
+
"""
|
32
|
+
|
33
|
+
def wrapper(wrapped, instance, args, kwargs):
|
34
|
+
"""
|
35
|
+
Wraps the 'generate' API call to add telemetry.
|
36
|
+
|
37
|
+
This collects metrics such as execution time, cost, and token usage, and handles errors
|
38
|
+
gracefully, adding details to the trace for observability.
|
39
|
+
|
40
|
+
Args:
|
41
|
+
wrapped: The original 'generate' method to be wrapped.
|
42
|
+
instance: The instance of the class where the original method is defined.
|
43
|
+
args: Positional arguments for the 'generate' method.
|
44
|
+
kwargs: Keyword arguments for the 'generate' method.
|
45
|
+
|
46
|
+
Returns:
|
47
|
+
The response from the original 'generate' method.
|
48
|
+
"""
|
49
|
+
|
50
|
+
# Check if streaming is enabled for the API call
|
51
|
+
streaming = kwargs.get("streaming", False)
|
52
|
+
|
53
|
+
# pylint: disable=no-else-return
|
54
|
+
if streaming:
|
55
|
+
# Special handling for streaming response to accommodate the nature of data flow
|
56
|
+
def stream_generator():
|
57
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
58
|
+
# Placeholder for aggregating streaming response
|
59
|
+
llmresponse = ""
|
60
|
+
|
61
|
+
# Loop through streaming events capturing relevant details
|
62
|
+
for chunk in wrapped(*args, **kwargs):
|
63
|
+
# Collect aggregated response from events
|
64
|
+
llmresponse += chunk
|
65
|
+
|
66
|
+
yield chunk
|
67
|
+
|
68
|
+
# Handling exception ensure observability without disrupting operation
|
69
|
+
try:
|
70
|
+
# Calculate cost of the operation
|
71
|
+
cost = 0
|
72
|
+
|
73
|
+
# pylint: disable=line-too-long
|
74
|
+
model = str(instance.model.model_path).rsplit('/', maxsplit=1)[-1] or "orca-mini-3b-gguf2-q4_0.gguf"
|
75
|
+
prompt = kwargs.get("prompt") or args[0] or ""
|
76
|
+
|
77
|
+
# Calculate cost of the operation
|
78
|
+
cost = 0
|
79
|
+
prompt_tokens = general_tokens(prompt)
|
80
|
+
completion_tokens = general_tokens(llmresponse)
|
81
|
+
total_tokens = prompt_tokens + completion_tokens
|
82
|
+
|
83
|
+
# Set base span attribues
|
84
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
85
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
86
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL)
|
87
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
88
|
+
SemanticConvetion.GEN_AI_TYPE_CHAT)
|
89
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
90
|
+
gen_ai_endpoint)
|
91
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
|
92
|
+
environment)
|
93
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
94
|
+
application_name)
|
95
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL,
|
96
|
+
model)
|
97
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_K,
|
98
|
+
kwargs.get("top_k", 40))
|
99
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_P,
|
100
|
+
kwargs.get("top_p", 0.4))
|
101
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MAX_TOKENS,
|
102
|
+
kwargs.get("max_tokens", 200))
|
103
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TEMPERATURE,
|
104
|
+
kwargs.get("temperature", 0.7))
|
105
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_FREQUENCY_PENALTY,
|
106
|
+
kwargs.get("frequency_penalty", 1.18))
|
107
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_IS_STREAM,
|
108
|
+
True)
|
109
|
+
if trace_content:
|
110
|
+
span.set_attribute(SemanticConvetion.GEN_AI_CONTENT_PROMPT,
|
111
|
+
prompt)
|
112
|
+
span.set_attribute(SemanticConvetion.GEN_AI_CONTENT_COMPLETION,
|
113
|
+
llmresponse)
|
114
|
+
|
115
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
|
116
|
+
prompt_tokens)
|
117
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COMPLETION_TOKENS,
|
118
|
+
completion_tokens)
|
119
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS,
|
120
|
+
total_tokens)
|
121
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST,
|
122
|
+
cost)
|
123
|
+
|
124
|
+
span.set_status(Status(StatusCode.OK))
|
125
|
+
|
126
|
+
if disable_metrics is False:
|
127
|
+
attributes = {
|
128
|
+
TELEMETRY_SDK_NAME:
|
129
|
+
"openlit",
|
130
|
+
SemanticConvetion.GEN_AI_APPLICATION_NAME:
|
131
|
+
application_name,
|
132
|
+
SemanticConvetion.GEN_AI_SYSTEM:
|
133
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL,
|
134
|
+
SemanticConvetion.GEN_AI_ENVIRONMENT:
|
135
|
+
environment,
|
136
|
+
SemanticConvetion.GEN_AI_TYPE:
|
137
|
+
SemanticConvetion.GEN_AI_TYPE_CHAT,
|
138
|
+
SemanticConvetion.GEN_AI_REQUEST_MODEL:
|
139
|
+
model
|
140
|
+
}
|
141
|
+
|
142
|
+
metrics["genai_requests"].add(1, attributes)
|
143
|
+
metrics["genai_total_tokens"].add(total_tokens, attributes)
|
144
|
+
metrics["genai_completion_tokens"].add(completion_tokens, attributes)
|
145
|
+
metrics["genai_prompt_tokens"].add(prompt_tokens, attributes)
|
146
|
+
metrics["genai_cost"].record(cost, attributes)
|
147
|
+
|
148
|
+
except Exception as e:
|
149
|
+
handle_exception(span, e)
|
150
|
+
logger.error("Error in trace creation: %s", e)
|
151
|
+
|
152
|
+
return stream_generator()
|
153
|
+
|
154
|
+
# Handling for non-streaming responses
|
155
|
+
else:
|
156
|
+
# pylint: disable=line-too-long
|
157
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
158
|
+
response = wrapped(*args, **kwargs)
|
159
|
+
|
160
|
+
# pylint: disable=line-too-long
|
161
|
+
model = str(instance.model.model_path).rsplit('/', maxsplit=1)[-1] or "orca-mini-3b-gguf2-q4_0.gguf"
|
162
|
+
prompt = kwargs.get("prompt") or args[0] or ""
|
163
|
+
|
164
|
+
# Calculate cost of the operation
|
165
|
+
cost = 0
|
166
|
+
prompt_tokens = general_tokens(prompt)
|
167
|
+
completion_tokens = general_tokens(response)
|
168
|
+
total_tokens = prompt_tokens + completion_tokens
|
169
|
+
|
170
|
+
try:
|
171
|
+
# Set base span attribues
|
172
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
173
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
174
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL)
|
175
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
176
|
+
SemanticConvetion.GEN_AI_TYPE_CHAT)
|
177
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
178
|
+
gen_ai_endpoint)
|
179
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
|
180
|
+
environment)
|
181
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
182
|
+
application_name)
|
183
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL,
|
184
|
+
model)
|
185
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_K,
|
186
|
+
kwargs.get("top_k", 40))
|
187
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_P,
|
188
|
+
kwargs.get("top_p", 0.4))
|
189
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MAX_TOKENS,
|
190
|
+
kwargs.get("max_tokens", 200))
|
191
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TEMPERATURE,
|
192
|
+
kwargs.get("temperature", 0.7))
|
193
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_FREQUENCY_PENALTY,
|
194
|
+
kwargs.get("frequency_penalty", 1.18))
|
195
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_IS_STREAM,
|
196
|
+
False)
|
197
|
+
if trace_content:
|
198
|
+
span.set_attribute(SemanticConvetion.GEN_AI_CONTENT_PROMPT,
|
199
|
+
prompt)
|
200
|
+
span.set_attribute(SemanticConvetion.GEN_AI_CONTENT_COMPLETION,
|
201
|
+
response)
|
202
|
+
|
203
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
|
204
|
+
prompt_tokens)
|
205
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COMPLETION_TOKENS,
|
206
|
+
completion_tokens)
|
207
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS,
|
208
|
+
total_tokens)
|
209
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST,
|
210
|
+
cost)
|
211
|
+
|
212
|
+
span.set_status(Status(StatusCode.OK))
|
213
|
+
|
214
|
+
if disable_metrics is False:
|
215
|
+
attributes = {
|
216
|
+
TELEMETRY_SDK_NAME:
|
217
|
+
"openlit",
|
218
|
+
SemanticConvetion.GEN_AI_APPLICATION_NAME:
|
219
|
+
application_name,
|
220
|
+
SemanticConvetion.GEN_AI_SYSTEM:
|
221
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL,
|
222
|
+
SemanticConvetion.GEN_AI_ENVIRONMENT:
|
223
|
+
environment,
|
224
|
+
SemanticConvetion.GEN_AI_TYPE:
|
225
|
+
SemanticConvetion.GEN_AI_TYPE_CHAT,
|
226
|
+
SemanticConvetion.GEN_AI_REQUEST_MODEL:
|
227
|
+
model
|
228
|
+
}
|
229
|
+
|
230
|
+
metrics["genai_requests"].add(1, attributes)
|
231
|
+
metrics["genai_total_tokens"].add(total_tokens, attributes)
|
232
|
+
metrics["genai_completion_tokens"].add(completion_tokens, attributes)
|
233
|
+
metrics["genai_prompt_tokens"].add(prompt_tokens, attributes)
|
234
|
+
metrics["genai_cost"].record(cost, attributes)
|
235
|
+
|
236
|
+
# Return original response
|
237
|
+
return response
|
238
|
+
|
239
|
+
except Exception as e:
|
240
|
+
handle_exception(span, e)
|
241
|
+
logger.error("Error in trace creation: %s", e)
|
242
|
+
|
243
|
+
# Return original response
|
244
|
+
return response
|
245
|
+
|
246
|
+
return wrapper
|
247
|
+
|
248
|
+
def embed(gen_ai_endpoint, version, environment, application_name,
|
249
|
+
tracer, pricing_info, trace_content, metrics, disable_metrics):
|
250
|
+
"""
|
251
|
+
Generates a telemetry wrapper for embeddings to collect metrics.
|
252
|
+
|
253
|
+
Args:
|
254
|
+
gen_ai_endpoint: Endpoint identifier for logging and tracing.
|
255
|
+
version: Version of the monitoring package.
|
256
|
+
environment: Deployment environment (e.g., production, staging).
|
257
|
+
application_name: Name of the application using the GPT4All API.
|
258
|
+
tracer: OpenTelemetry tracer for creating spans.
|
259
|
+
pricing_info: Information used for calculating the cost of GPT4All usage.
|
260
|
+
trace_content: Flag indicating whether to trace the actual content.
|
261
|
+
|
262
|
+
Returns:
|
263
|
+
A function that wraps the embeddings method to add telemetry.
|
264
|
+
"""
|
265
|
+
|
266
|
+
def wrapper(wrapped, instance, args, kwargs):
|
267
|
+
"""
|
268
|
+
Wraps the 'embeddings' API call to add telemetry.
|
269
|
+
|
270
|
+
This collects metrics such as execution time, cost, and token usage, and handles errors
|
271
|
+
gracefully, adding details to the trace for observability.
|
272
|
+
|
273
|
+
Args:
|
274
|
+
wrapped: The original 'embeddings' method to be wrapped.
|
275
|
+
instance: The instance of the class where the original method is defined.
|
276
|
+
args: Positional arguments for the 'embeddings' method.
|
277
|
+
kwargs: Keyword arguments for the 'embeddings' method.
|
278
|
+
|
279
|
+
Returns:
|
280
|
+
The response from the original 'embeddings' method.
|
281
|
+
"""
|
282
|
+
|
283
|
+
with tracer.start_as_current_span(gen_ai_endpoint, kind= SpanKind.CLIENT) as span:
|
284
|
+
response = wrapped(*args, **kwargs)
|
285
|
+
|
286
|
+
try:
|
287
|
+
# pylint: disable=line-too-long
|
288
|
+
model = str(instance.gpt4all.model.model_path).rsplit('/', maxsplit=1)[-1] or "all-MiniLM-L6-v2.gguf2.f16.gguf"
|
289
|
+
prompt = kwargs.get("prompt") or args[0] or ""
|
290
|
+
|
291
|
+
# Calculate cost of the operation
|
292
|
+
cost = 0
|
293
|
+
prompt_tokens = general_tokens(prompt)
|
294
|
+
|
295
|
+
# Set Span attributes
|
296
|
+
span.set_attribute(TELEMETRY_SDK_NAME, "openlit")
|
297
|
+
span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM,
|
298
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL)
|
299
|
+
span.set_attribute(SemanticConvetion.GEN_AI_TYPE,
|
300
|
+
SemanticConvetion.GEN_AI_TYPE_EMBEDDING)
|
301
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENDPOINT,
|
302
|
+
gen_ai_endpoint)
|
303
|
+
span.set_attribute(SemanticConvetion.GEN_AI_ENVIRONMENT,
|
304
|
+
environment)
|
305
|
+
span.set_attribute(SemanticConvetion.GEN_AI_APPLICATION_NAME,
|
306
|
+
application_name)
|
307
|
+
span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL,
|
308
|
+
model)
|
309
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_PROMPT_TOKENS,
|
310
|
+
prompt_tokens)
|
311
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_TOTAL_TOKENS,
|
312
|
+
prompt_tokens)
|
313
|
+
span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST,
|
314
|
+
cost)
|
315
|
+
if trace_content:
|
316
|
+
span.set_attribute(SemanticConvetion.GEN_AI_CONTENT_PROMPT,
|
317
|
+
prompt)
|
318
|
+
|
319
|
+
span.set_status(Status(StatusCode.OK))
|
320
|
+
|
321
|
+
if disable_metrics is False:
|
322
|
+
attributes = {
|
323
|
+
TELEMETRY_SDK_NAME:
|
324
|
+
"openlit",
|
325
|
+
SemanticConvetion.GEN_AI_APPLICATION_NAME:
|
326
|
+
application_name,
|
327
|
+
SemanticConvetion.GEN_AI_SYSTEM:
|
328
|
+
SemanticConvetion.GEN_AI_SYSTEM_GPT4ALL,
|
329
|
+
SemanticConvetion.GEN_AI_ENVIRONMENT:
|
330
|
+
environment,
|
331
|
+
SemanticConvetion.GEN_AI_TYPE:
|
332
|
+
SemanticConvetion.GEN_AI_TYPE_EMBEDDING,
|
333
|
+
SemanticConvetion.GEN_AI_REQUEST_MODEL:
|
334
|
+
model
|
335
|
+
}
|
336
|
+
|
337
|
+
metrics["genai_requests"].add(1, attributes)
|
338
|
+
metrics["genai_total_tokens"].add(prompt_tokens, attributes)
|
339
|
+
metrics["genai_prompt_tokens"].add(prompt_tokens, attributes)
|
340
|
+
metrics["genai_cost"].record(cost, attributes)
|
341
|
+
|
342
|
+
# Return original response
|
343
|
+
return response
|
344
|
+
|
345
|
+
except Exception as e:
|
346
|
+
handle_exception(span, e)
|
347
|
+
logger.error("Error in trace creation: %s", e)
|
348
|
+
|
349
|
+
# Return original response
|
350
|
+
return response
|
351
|
+
|
352
|
+
return wrapper
|
@@ -25,7 +25,7 @@ WRAPPED_METHODS = [
|
|
25
25
|
]
|
26
26
|
|
27
27
|
class LlamaIndexInstrumentor(BaseInstrumentor):
|
28
|
-
"""An instrumentor for
|
28
|
+
"""An instrumentor for LlamaIndex's client library."""
|
29
29
|
|
30
30
|
def instrumentation_dependencies(self) -> Collection[str]:
|
31
31
|
return _instruments
|
@@ -119,9 +119,9 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
119
119
|
span.set_attribute(SemanticConvetion.DB_OPERATION_STATUS,
|
120
120
|
response.status)
|
121
121
|
span.set_attribute(SemanticConvetion.DB_VECTOR_COUNT,
|
122
|
-
object_count(kwargs.get("points")))
|
122
|
+
object_count(kwargs.get("points", [])))
|
123
123
|
span.set_attribute(SemanticConvetion.DB_PAYLOAD_COUNT,
|
124
|
-
object_count(kwargs.get("payload")))
|
124
|
+
object_count(kwargs.get("payload", [])))
|
125
125
|
|
126
126
|
elif gen_ai_endpoint == "qdrant.retrieve":
|
127
127
|
db_operation = SemanticConvetion.DB_OPERATION_QUERY
|
@@ -130,7 +130,7 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
130
130
|
span.set_attribute(SemanticConvetion.DB_COLLECTION_NAME,
|
131
131
|
kwargs.get("collection_name", ""))
|
132
132
|
span.set_attribute(SemanticConvetion.DB_STATEMENT,
|
133
|
-
str(kwargs.get("ids")))
|
133
|
+
str(kwargs.get("ids", "")))
|
134
134
|
|
135
135
|
elif gen_ai_endpoint == "qdrant.scroll":
|
136
136
|
db_operation = SemanticConvetion.DB_OPERATION_QUERY
|
@@ -139,7 +139,7 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
139
139
|
span.set_attribute(SemanticConvetion.DB_COLLECTION_NAME,
|
140
140
|
kwargs.get("collection_name", ""))
|
141
141
|
span.set_attribute(SemanticConvetion.DB_STATEMENT,
|
142
|
-
str(kwargs.get("scroll_filter")))
|
142
|
+
str(kwargs.get("scroll_filter", "")))
|
143
143
|
|
144
144
|
elif gen_ai_endpoint in ["qdrant.search", "qdrant.search_groups"]:
|
145
145
|
db_operation = SemanticConvetion.DB_OPERATION_QUERY
|
@@ -148,7 +148,7 @@ def general_wrap(gen_ai_endpoint, version, environment, application_name,
|
|
148
148
|
span.set_attribute(SemanticConvetion.DB_COLLECTION_NAME,
|
149
149
|
kwargs.get("collection_name", ""))
|
150
150
|
span.set_attribute(SemanticConvetion.DB_STATEMENT,
|
151
|
-
str(kwargs.get("query_vector")))
|
151
|
+
str(kwargs.get("query_vector", "")))
|
152
152
|
|
153
153
|
elif gen_ai_endpoint == "qdrant.recommend":
|
154
154
|
db_operation = SemanticConvetion.DB_OPERATION_QUERY
|
@@ -26,6 +26,7 @@ class SemanticConvetion:
|
|
26
26
|
GEN_AI_HUB_REPO = "gen_ai.hub.repo"
|
27
27
|
GEN_AI_RETRIEVAL_SOURCE = "gen_ai.retrieval.source"
|
28
28
|
GEN_AI_REQUESTS = "gen_ai.total.requests"
|
29
|
+
GEN_AI_DATA_SOURCES = "gen_ai.data_source_count"
|
29
30
|
|
30
31
|
# GenAI Request
|
31
32
|
GEN_AI_REQUEST_MODEL = "gen_ai.request.model"
|
@@ -71,6 +72,11 @@ class SemanticConvetion:
|
|
71
72
|
GEN_AI_CONTENT_COMPLETION = "gen_ai.content.completion"
|
72
73
|
GEN_AI_CONTENT_REVISED_PROMPT = "gen_ai.content.revised_prompt"
|
73
74
|
|
75
|
+
# GenAI Evaluation Metrics
|
76
|
+
GEN_AI_EVAL_CONTEXT_RELEVANCY = "gen_ai.eval.context_relevancy"
|
77
|
+
GEN_AI_EVAL_ANSWER_RELEVANCY = "gen_ai.eval.answer_relevancy"
|
78
|
+
GEN_AI_EVAL_GROUNDEDNESS = "gen_ai.eval.groundedness"
|
79
|
+
|
74
80
|
GEN_AI_TYPE_CHAT = "chat"
|
75
81
|
GEN_AI_TYPE_EMBEDDING = "embedding"
|
76
82
|
GEN_AI_TYPE_IMAGE = "image"
|
@@ -89,9 +95,11 @@ class SemanticConvetion:
|
|
89
95
|
GEN_AI_SYSTEM_VERTEXAI = "vertexai"
|
90
96
|
GEN_AI_SYSTEM_GROQ = "groq"
|
91
97
|
GEN_AI_SYSTEM_OLLAMA = "ollama"
|
98
|
+
GEN_AI_SYSTEM_GPT4ALL = "gpt4all"
|
92
99
|
GEN_AI_SYSTEM_LANGCHAIN = "langchain"
|
93
100
|
GEN_AI_SYSTEM_LLAMAINDEX = "llama_index"
|
94
101
|
GEN_AI_SYSTEM_HAYSTACK = "haystack"
|
102
|
+
GEN_AI_SYSTEM_EMBEDCHAIN = "embedchain"
|
95
103
|
|
96
104
|
# Vector DB
|
97
105
|
DB_REQUESTS = "db.total.requests"
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|
File without changes
|