improvements to exception mapping

docs/my-website/docs/exception_mapping.md

@@ -43,13 +43,14 @@ Base case - we return the original exception.
 |               | ContextWindowExceededError | AuthenticationError | InvalidRequestError | RateLimitError | ServiceUnavailableError |
 |---------------|----------------------------|---------------------|---------------------|---------------|-------------------------|
 | Anthropic     | ✅                          | ✅                   | ✅                   | ✅             |                         |
-| OpenAI        | ✅                          | ✅                     |✅                     |✅               |✅                         |
+| OpenAI        | ✅                          | ✅                     |✅                     |✅               |✅|
 | Replicate     | ✅                          | ✅                   | ✅                   | ✅             | ✅                       |
-| Cohere        | ✅                          | ✅                   |                     | ✅             |                         |
+| Cohere        | ✅                          | ✅                   | ✅                    | ✅             | ✅                        |
 | Huggingface   | ✅                          | ✅                   | ✅                   | ✅             |                         |
-| Openrouter    | ✅                          | ✅                   |                     | ✅             |                         |
+| Openrouter    | ✅                          | ✅                   | ✅                    | ✅             |                         |
 | AI21          | ✅                          | ✅                   | ✅                   | ✅             |                         |
 | TogetherAI    | ✅                          | ✅                   | ✅                   | ✅             |                         |
+| AlephAlpha    | ✅                          | ✅                   | ✅                   | ✅             | ✅                        |
 
 
 > For a deeper understanding of these exceptions, you can check out [this](https://github.com/BerriAI/litellm/blob/d7e58d13bf9ba9edbab2ab2f096f3de7547f35fa/litellm/utils.py#L1544) implementation for additional insights.

litellm/tests/test_exceptions.py

@@ -12,7 +12,7 @@ from litellm import (
     embedding,
     completion,
 #     AuthenticationError,
-#     InvalidRequestError,
+    InvalidRequestError,
     ContextWindowExceededError,
 #     RateLimitError,
 #     ServiceUnavailableError,
@@ -34,8 +34,7 @@ litellm.vertex_location = "us-central1"
 
 # Approach: Run each model through the test -> assert if the correct error (always the same one) is triggered
 
-models = ["gpt-3.5-turbo"]
+models = ["command-nightly"]
-test_model = "claude-instant-1"
 
 # Test 1: Context Window Errors 
 @pytest.mark.parametrize("model", models)
@@ -73,6 +72,9 @@ def invalid_auth(model):  # set the model key to an invalid key, depending on th
         elif model in litellm.openrouter_models:
             temporary_key = os.environ["OPENROUTER_API_KEY"]
             os.environ["OPENROUTER_API_KEY"] = "bad-key"
+        elif model in litellm.aleph_alpha_models:
+            temporary_key = os.environ["ALEPH_ALPHA_API_KEY"]
+            os.environ["ALEPH_ALPHA_API_KEY"] = "bad-key"
         elif (
             model
             == "replicate/llama-2-70b-chat:2c1608e18606fad2812020dc541930f2d0495ce32eee50074220b87300bc16e1"
@@ -115,8 +117,18 @@ def invalid_auth(model):  # set the model key to an invalid key, depending on th
             os.environ["AI21_API_KEY"] = temporary_key
         elif ("togethercomputer" in model):
             os.environ["TOGETHERAI_API_KEY"] = temporary_key
+        elif model in litellm.aleph_alpha_models:
+            os.environ["ALEPH_ALPHA_API_KEY"] = temporary_key
     return
 
+# Test 3: Invalid Request Error 
+@pytest.mark.parametrize("model", models)
+def test_invalid_request_error(model):
+    messages = [{"content": "hey, how's it going?", "role": "user"}]
+
+    with pytest.raises(InvalidRequestError):
+        completion(model=model, messages=messages, max_tokens="hello world")
+
 # Test 3: Rate Limit Errors
 # def test_model_call(model):
 #     try:

litellm/utils.py

@@ -1786,6 +1786,20 @@ def exception_type(model, original_exception, custom_llm_provider):
                         llm_provider="cohere",
                         model=model
                     )
+                elif "invalid type:" in error_str:
+                    exception_mapping_worked = True
+                    raise InvalidRequestError(
+                        message=f"CohereException - {original_exception.message}",
+                        llm_provider="cohere",
+                        model=model
+                    )
+                elif "Unexpected server error" in error_str:
+                    exception_mapping_worked = True
+                    raise ServiceUnavailableError(
+                        message=f"CohereException - {original_exception.message}",
+                        llm_provider="cohere",
+                        model=model
+                    )
                 else:
                     if hasattr(original_exception, "status_code"):
                         exception_mapping_worked = True
@@ -1938,6 +1952,46 @@ def exception_type(model, original_exception, custom_llm_provider):
                         llm_provider="together_ai",
                         model=model
                     )
+            elif model in litellm.aleph_alpha_models:
+                if "This is longer than the model's maximum context length" in error_str:
+                    exception_mapping_worked = True
+                    raise ContextWindowExceededError(
+                        message=f"AlephAlphaException - {original_exception.message}",
+                        llm_provider="aleph_alpha", 
+                        model=model
+                    )
+                elif hasattr(original_exception, "status_code"):
+                    print(f"status code: {original_exception.status_code}")
+                    if original_exception.status_code == 401:
+                        exception_mapping_worked = True
+                        raise AuthenticationError(
+                            message=f"AlephAlphaException - {original_exception.message}",
+                            llm_provider="aleph_alpha",
+                            model=model
+                        )
+                    elif original_exception.status_code == 400:
+                        exception_mapping_worked = True
+                        raise InvalidRequestError(
+                            message=f"AlephAlphaException - {original_exception.message}",
+                            llm_provider="aleph_alpha",
+                            model=model
+                        )
+                    elif original_exception.status_code == 429:
+                        exception_mapping_worked = True
+                        raise RateLimitError(
+                            message=f"AlephAlphaException - {original_exception.message}",
+                            llm_provider="aleph_alpha",
+                            model=model
+                        )
+                    elif original_exception.status_code == 500:
+                        exception_mapping_worked = True
+                        raise ServiceUnavailableError(
+                            message=f"AlephAlphaException - {original_exception.message}",
+                            llm_provider="aleph_alpha",
+                            model=model
+                        )
+                    raise original_exception
+                raise original_exception
             elif custom_llm_provider == "vllm":
                 if hasattr(original_exception, "status_code"):
                     if original_exception.status_code == 0: