Merge pull request #268 from krrishdholakia/patch-1

Add support for Bedrock, Huggingface, Replicate, TogetherAI, Azure, Anthropic, etc.

Author: penguine-ip

deepeval/metrics/llm_eval_metric.py

@@ -8,7 +8,7 @@
 from deepeval.types import LLMTestCaseParams
 from deepeval.chat_completion.retry import call_openai_with_retry
 from pydantic import BaseModel
-import openai
+from litellm import completion, text_completion
 
 
 class LLMEvalMetricResponse(BaseModel):
@@ -65,7 +65,7 @@ def generate_evaluation_steps(self):
         prompt: dict = evaluation_steps_template.format(criteria=self.criteria)
 
         res = call_openai_with_retry(
-            lambda: openai.ChatCompletion.create(
+            lambda: completion(
                 model=self.model,
                 messages=[
                     {
@@ -92,7 +92,7 @@ def evaluate(self, test_case: LLMTestCase):
         )
 
         res = call_openai_with_retry(
-            lambda: openai.ChatCompletion.create(
+            lambda: completion(
                 model=self.model,
                 messages=[{"role": "system", "content": prompt}],
                 max_tokens=5,