fix OTEL semantic conventions does not exist

litellm/integrations/opentelemetry.py

@@ -247,7 +247,7 @@ class OpenTelemetry(CustomLogger):
         span.end(end_time=self._to_ns(end_time))
 
     def set_tools_attributes(self, span: Span, tools):
-        from opentelemetry.semconv.ai import SpanAttributes
+        from litellm.proxy._types import SpanAttributes
         import json
 
         if not tools:
@@ -272,7 +272,7 @@ class OpenTelemetry(CustomLogger):
             pass
 
     def set_attributes(self, span: Span, kwargs, response_obj):
-        from opentelemetry.semconv.ai import SpanAttributes
+        from litellm.proxy._types import SpanAttributes
 
         optional_params = kwargs.get("optional_params", {})
         litellm_params = kwargs.get("litellm_params", {}) or {}
@@ -407,7 +407,7 @@ class OpenTelemetry(CustomLogger):
             )
 
     def set_raw_request_attributes(self, span: Span, kwargs, response_obj):
-        from opentelemetry.semconv.ai import SpanAttributes
+        from litellm.proxy._types import SpanAttributes
 
         optional_params = kwargs.get("optional_params", {})
         litellm_params = kwargs.get("litellm_params", {}) or {}

litellm/proxy/_types.py

@@ -1513,3 +1513,51 @@ class SpendLogsPayload(TypedDict):
     request_tags: str  # json str
     team_id: Optional[str]
     end_user: Optional[str]
+
+
+class SpanAttributes(str, enum.Enum):
+    # Note: We've taken this from opentelemetry-semantic-conventions-ai
+    # I chose to not add a new dependency to litellm for this
+
+    # Semantic Conventions for LLM requests, this needs to be removed after
+    # OpenTelemetry Semantic Conventions support Gen AI.
+    # Issue at https://github.com/open-telemetry/opentelemetry-python/issues/3868
+    # Refer to https://github.com/open-telemetry/semantic-conventions/blob/main/docs/gen-ai/llm-spans.md
+
+    LLM_SYSTEM = "gen_ai.system"
+    LLM_REQUEST_MODEL = "gen_ai.request.model"
+    LLM_REQUEST_MAX_TOKENS = "gen_ai.request.max_tokens"
+    LLM_REQUEST_TEMPERATURE = "gen_ai.request.temperature"
+    LLM_REQUEST_TOP_P = "gen_ai.request.top_p"
+    LLM_PROMPTS = "gen_ai.prompt"
+    LLM_COMPLETIONS = "gen_ai.completion"
+    LLM_RESPONSE_MODEL = "gen_ai.response.model"
+    LLM_USAGE_COMPLETION_TOKENS = "gen_ai.usage.completion_tokens"
+    LLM_USAGE_PROMPT_TOKENS = "gen_ai.usage.prompt_tokens"
+    LLM_TOKEN_TYPE = "gen_ai.token.type"
+    # To be added
+    # LLM_RESPONSE_FINISH_REASON = "gen_ai.response.finish_reasons"
+    # LLM_RESPONSE_ID = "gen_ai.response.id"
+
+    # LLM
+    LLM_REQUEST_TYPE = "llm.request.type"
+    LLM_USAGE_TOTAL_TOKENS = "llm.usage.total_tokens"
+    LLM_USAGE_TOKEN_TYPE = "llm.usage.token_type"
+    LLM_USER = "llm.user"
+    LLM_HEADERS = "llm.headers"
+    LLM_TOP_K = "llm.top_k"
+    LLM_IS_STREAMING = "llm.is_streaming"
+    LLM_FREQUENCY_PENALTY = "llm.frequency_penalty"
+    LLM_PRESENCE_PENALTY = "llm.presence_penalty"
+    LLM_CHAT_STOP_SEQUENCES = "llm.chat.stop_sequences"
+    LLM_REQUEST_FUNCTIONS = "llm.request.functions"
+    LLM_REQUEST_REPETITION_PENALTY = "llm.request.repetition_penalty"
+    LLM_RESPONSE_FINISH_REASON = "llm.response.finish_reason"
+    LLM_RESPONSE_STOP_REASON = "llm.response.stop_reason"
+    LLM_CONTENT_COMPLETION_CHUNK = "llm.content.completion.chunk"
+
+    # OpenAI
+    LLM_OPENAI_RESPONSE_SYSTEM_FINGERPRINT = "gen_ai.openai.system_fingerprint"
+    LLM_OPENAI_API_BASE = "gen_ai.openai.api_base"
+    LLM_OPENAI_API_VERSION = "gen_ai.openai.api_version"
+    LLM_OPENAI_API_TYPE = "gen_ai.openai.api_type"

requirements.txt

@@ -31,7 +31,6 @@ azure-identity==1.15.0 # for azure content safety
 opentelemetry-api==1.25.0
 opentelemetry-sdk==1.25.0
 opentelemetry-exporter-otlp==1.25.0
-opentelemetry-semantic-conventions-ai==0.3.1
 
 ### LITELLM PACKAGE DEPENDENCIES
 python-dotenv==1.0.0 # for env