diff --git a/litellm/caching.py b/litellm/caching.py index 1d993927b..556ab4fb5 100644 --- a/litellm/caching.py +++ b/litellm/caching.py @@ -222,7 +222,14 @@ class Cache: Returns: str: The cache key generated from the arguments, or None if no cache key could be generated. """ - cache_key ="" + cache_key = "" + print_verbose(f"\nGetting Cache key. Kwargs: {kwargs}") + + # for streaming, we use preset_cache_key. It's created in wrapper(), we do this because optional params like max_tokens, get transformed for bedrock -> max_new_tokens + if kwargs.get("litellm_params", {}).get("preset_cache_key", None) is not None: + print_verbose(f"\nReturning preset cache key: {cache_key}") + return kwargs.get("litellm_params", {}).get("preset_cache_key", None) + # sort kwargs by keys, since model: [gpt-4, temperature: 0.2, max_tokens: 200] == [temperature: 0.2, max_tokens: 200, model: gpt-4] completion_kwargs = ["model", "messages", "temperature", "top_p", "n", "stop", "max_tokens", "presence_penalty", "frequency_penalty", "logit_bias", "user", "response_format", "seed", "tools", "tool_choice"] for param in completion_kwargs: @@ -245,6 +252,7 @@ class Cache: continue # ignore None params param_value = kwargs[param] cache_key+= f"{str(param)}: {str(param_value)}" + print_verbose(f"\nCreated cache key: {cache_key}") return cache_key def generate_streaming_content(self, content): diff --git a/litellm/main.py b/litellm/main.py index f97659a32..e6642d1d1 100644 --- a/litellm/main.py +++ b/litellm/main.py @@ -344,13 +344,14 @@ def completion( final_prompt_value = kwargs.get("final_prompt_value", None) bos_token = kwargs.get("bos_token", None) eos_token = kwargs.get("eos_token", None) + preset_cache_key = kwargs.get("preset_cache_key", None) hf_model_name = kwargs.get("hf_model_name", None) ### ASYNC CALLS ### acompletion = kwargs.get("acompletion", False) client = kwargs.get("client", None) ######## end of unpacking kwargs ########### openai_params = ["functions", "function_call", "temperature", "temperature", "top_p", "n", "stream", "stop", "max_tokens", "presence_penalty", "frequency_penalty", "logit_bias", "user", "request_timeout", "api_base", "api_version", "api_key", "deployment_id", "organization", "base_url", "default_headers", "timeout", "response_format", "seed", "tools", "tool_choice", "max_retries"] - litellm_params = ["metadata", "acompletion", "caching", "return_async", "mock_response", "api_key", "api_version", "api_base", "force_timeout", "logger_fn", "verbose", "custom_llm_provider", "litellm_logging_obj", "litellm_call_id", "use_client", "id", "fallbacks", "azure", "headers", "model_list", "num_retries", "context_window_fallback_dict", "roles", "final_prompt_value", "bos_token", "eos_token", "request_timeout", "complete_response", "self", "client", "rpm", "tpm", "input_cost_per_token", "output_cost_per_token", "hf_model_name", "model_info", "proxy_server_request"] + litellm_params = ["metadata", "acompletion", "caching", "return_async", "mock_response", "api_key", "api_version", "api_base", "force_timeout", "logger_fn", "verbose", "custom_llm_provider", "litellm_logging_obj", "litellm_call_id", "use_client", "id", "fallbacks", "azure", "headers", "model_list", "num_retries", "context_window_fallback_dict", "roles", "final_prompt_value", "bos_token", "eos_token", "request_timeout", "complete_response", "self", "client", "rpm", "tpm", "input_cost_per_token", "output_cost_per_token", "hf_model_name", "model_info", "proxy_server_request", "preset_cache_key"] default_params = openai_params + litellm_params non_default_params = {k: v for k,v in kwargs.items() if k not in default_params} # model-specific params - pass them straight to the model/provider if mock_response: @@ -460,7 +461,8 @@ def completion( completion_call_id=id, metadata=metadata, model_info=model_info, - proxy_server_request=proxy_server_request + proxy_server_request=proxy_server_request, + preset_cache_key=preset_cache_key ) logging.update_environment_variables(model=model, user=user, optional_params=optional_params, litellm_params=litellm_params) if custom_llm_provider == "azure": @@ -1784,7 +1786,7 @@ def embedding( proxy_server_request = kwargs.get("proxy_server_request", None) aembedding = kwargs.pop("aembedding", None) openai_params = ["functions", "function_call", "temperature", "temperature", "top_p", "n", "stream", "stop", "max_tokens", "presence_penalty", "frequency_penalty", "logit_bias", "user", "request_timeout", "api_base", "api_version", "api_key", "deployment_id", "organization", "base_url", "default_headers", "timeout", "response_format", "seed", "tools", "tool_choice", "max_retries", "encoding_format"] - litellm_params = ["metadata", "acompletion", "caching", "return_async", "mock_response", "api_key", "api_version", "api_base", "force_timeout", "logger_fn", "verbose", "custom_llm_provider", "litellm_logging_obj", "litellm_call_id", "use_client", "id", "fallbacks", "azure", "headers", "model_list", "num_retries", "context_window_fallback_dict", "roles", "final_prompt_value", "bos_token", "eos_token", "request_timeout", "complete_response", "self", "client", "rpm", "tpm", "input_cost_per_token", "output_cost_per_token", "hf_model_name", "proxy_server_request", "model_info"] + litellm_params = ["metadata", "acompletion", "caching", "return_async", "mock_response", "api_key", "api_version", "api_base", "force_timeout", "logger_fn", "verbose", "custom_llm_provider", "litellm_logging_obj", "litellm_call_id", "use_client", "id", "fallbacks", "azure", "headers", "model_list", "num_retries", "context_window_fallback_dict", "roles", "final_prompt_value", "bos_token", "eos_token", "request_timeout", "complete_response", "self", "client", "rpm", "tpm", "input_cost_per_token", "output_cost_per_token", "hf_model_name", "proxy_server_request", "model_info", "preset_cache_key"] default_params = openai_params + litellm_params non_default_params = {k: v for k,v in kwargs.items() if k not in default_params} # model-specific params - pass them straight to the model/provider optional_params = {}