Fix/oai parameters 1552 (#1557)

* support temperature between 0 and 2

* added parsing of presence_penalty from model parameters. OAI endpoint now correctly uses (frequency penalty=parameters.repetition penalty) and (presence penalty=parameters.presence_penalty) for both chat and completion paths.

* added parsing of presence_penalty from model parameters. OAI endpoint now correctly uses (frequency penalty=parameters.repetition penalty) and (presence penalty=parameters.presence_penalty) for both chat and completion paths.

src/lib/server/endpoints/openai/endpointOai.ts

@@ -167,6 +167,7 @@ export async function endpointOai(
 				temperature: parameters?.temperature,
 				top_p: parameters?.top_p,
 				frequency_penalty: parameters?.repetition_penalty,
+				presence_penalty: parameters?.presence_penalty,
 			};
 
 			const openAICompletion = await openai.completions.create(body, {
@@ -245,6 +246,7 @@ export async function endpointOai(
 				temperature: parameters?.temperature,
 				top_p: parameters?.top_p,
 				frequency_penalty: parameters?.repetition_penalty,
+				presence_penalty: parameters?.presence_penalty,
 				...(toolCallChoices.length > 0 ? { tools: toolCallChoices, tool_choice: "auto" } : {}),
 			};
 

src/lib/server/models.ts

@@ -52,13 +52,14 @@ const modelConfig = z.object({
 	endpoints: z.array(endpointSchema).optional(),
 	parameters: z
 		.object({
-			temperature: z.number().min(0).max(1).optional(),
+			temperature: z.number().min(0).max(2).optional(),
 			truncate: z.number().int().positive().optional(),
 			max_new_tokens: z.number().int().positive().optional(),
 			stop: z.array(z.string()).optional(),
 			top_p: z.number().positive().optional(),
 			top_k: z.number().positive().optional(),
 			repetition_penalty: z.number().min(-2).max(2).optional(),
+			presence_penalty: z.number().min(-2).max(2).optional(),
 		})
 		.passthrough()
 		.optional(),