fix: add missing seed parameter to ollama input

Current ollama interfacing does not allow for seed, which is supported in https://github.com/ollama/ollama/blob/main/docs/api.md#parameters and https://github.com/ollama/ollama/blob/main/docs/modelfile.md#valid-parameters-and-values

This resolves that by adding in handling of seed parameter.

litellm/llms/ollama.py

@@ -45,6 +45,8 @@ class OllamaConfig:
 
     - `temperature` (float): The temperature of the model. Increasing the temperature will make the model answer more creatively. Default: 0.8. Example usage: temperature 0.7
 
+    - `seed` (int): Sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt. Example usage: seed 42
+
     - `stop` (string[]): Sets the stop sequences to use. Example usage: stop "AI assistant:"
 
     - `tfs_z` (float): Tail free sampling is used to reduce the impact of less probable tokens from the output. A higher value (e.g., 2.0) will reduce the impact more, while a value of 1.0 disables this setting. Default: 1. Example usage: tfs_z 1
@@ -69,6 +71,7 @@ class OllamaConfig:
     repeat_last_n: Optional[int] = None
     repeat_penalty: Optional[float] = None
     temperature: Optional[float] = None
+    seed: Optional[int] = None
     stop: Optional[list] = (
         None  # stop is a list based on this - https://github.com/ollama/ollama/pull/442
     )
@@ -90,6 +93,7 @@ class OllamaConfig:
         repeat_last_n: Optional[int] = None,
         repeat_penalty: Optional[float] = None,
         temperature: Optional[float] = None,
+        seed: Optional[int] = None,
         stop: Optional[list] = None,
         tfs_z: Optional[float] = None,
         num_predict: Optional[int] = None,
@@ -120,6 +124,19 @@ class OllamaConfig:
             )
             and v is not None
         }
+    def get_supported_openai_params(
+        self,
+    ):
+        return [
+            "max_tokens",
+            "stream",
+            "top_p",
+            "temperature",
+            "seed",
+            "frequency_penalty",
+            "stop",
+            "response_format",
+        ]
 
 # ollama wants plain base64 jpeg/png files as images.  strip any leading dataURI
 # and convert to jpeg if necessary.

litellm/llms/ollama_chat.py

@@ -45,6 +45,8 @@ class OllamaChatConfig:
 
     - `temperature` (float): The temperature of the model. Increasing the temperature will make the model answer more creatively. Default: 0.8. Example usage: temperature 0.7
 
+    - `seed` (int): Sets the random number seed to use for generation. Setting this to a specific number will make the model generate the same text for the same prompt. Example usage: seed 42
+
     - `stop` (string[]): Sets the stop sequences to use. Example usage: stop "AI assistant:"
 
     - `tfs_z` (float): Tail free sampling is used to reduce the impact of less probable tokens from the output. A higher value (e.g., 2.0) will reduce the impact more, while a value of 1.0 disables this setting. Default: 1. Example usage: tfs_z 1
@@ -69,6 +71,7 @@ class OllamaChatConfig:
     repeat_last_n: Optional[int] = None
     repeat_penalty: Optional[float] = None
     temperature: Optional[float] = None
+    seed: Optional[int] = None
     stop: Optional[list] = (
         None  # stop is a list based on this - https://github.com/ollama/ollama/pull/442
     )
@@ -90,6 +93,7 @@ class OllamaChatConfig:
         repeat_last_n: Optional[int] = None,
         repeat_penalty: Optional[float] = None,
         temperature: Optional[float] = None,
+        seed: Optional[int] = None,
         stop: Optional[list] = None,
         tfs_z: Optional[float] = None,
         num_predict: Optional[int] = None,
@@ -130,6 +134,7 @@ class OllamaChatConfig:
             "stream",
             "top_p",
             "temperature",
+            "seed",
             "frequency_penalty",
             "stop",
             "tools",
@@ -146,6 +151,8 @@ class OllamaChatConfig:
                 optional_params["stream"] = value
             if param == "temperature":
                 optional_params["temperature"] = value
+            if param == "seed":
+                optional_params["seed"] = value
             if param == "top_p":
                 optional_params["top_p"] = value
             if param == "frequency_penalty":

litellm/utils.py

@@ -5752,6 +5752,8 @@ def get_optional_params(
             optional_params["stream"] = stream
         if temperature is not None:
             optional_params["temperature"] = temperature
+        if seed is not None:
+            optional_params["seed"] = seed
         if top_p is not None:
             optional_params["top_p"] = top_p
         if frequency_penalty is not None:
@@ -6404,6 +6406,8 @@ def get_supported_openai_params(
             return ["stream", "temperature", "max_tokens"]
         elif model.startswith("mistral"):
             return ["max_tokens", "temperature", "stop", "top_p", "stream"]
+    elif custom_llm_provider == "ollama":
+        return litellm.OllamaConfig().get_supported_openai_params()
     elif custom_llm_provider == "ollama_chat":
         return litellm.OllamaChatConfig().get_supported_openai_params()
     elif custom_llm_provider == "anthropic":
@@ -6573,16 +6577,6 @@ def get_supported_openai_params(
         ]
     elif custom_llm_provider == "cloudflare":
         return ["max_tokens", "stream"]
-    elif custom_llm_provider == "ollama":
-        return [
-            "max_tokens",
-            "stream",
-            "top_p",
-            "temperature",
-            "frequency_penalty",
-            "stop",
-            "response_format",
-        ]
     elif custom_llm_provider == "nlp_cloud":
         return [
             "max_tokens",