server: add logprobs and top_logprobs support to Ollama's API (#12899)

Adds logprobs support to Ollama's API including support for Ollama's
OpenAI-compatible API. By specifying the new 'logprobs' boolean parameter
in the API, Ollama will return the log probabilities for each token generated.
'top_logprobs', an integer value can also be specified up to the value 20.
When specified, the API will also provide the number of most likely tokens to
return at each token position

Co-authored-by: Baptiste Jamin <baptiste@crisp.chat>

llama/llama.go

@@ -217,6 +217,19 @@ func (c *Context) GetEmbeddingsIth(i int) []float32 {
 	return embeddings
 }
 
+// GetLogitsIth gets the logits for the ith token
+func (c *Context) GetLogitsIth(i int) []float32 {
+	logits := unsafe.Pointer(C.llama_get_logits_ith(c.c, C.int32_t(i)))
+	if logits == nil {
+		return nil
+	}
+
+	vocabSize := c.Model().NumVocab()
+	result := make([]float32, vocabSize)
+	_ = copy(result, unsafe.Slice((*float32)(logits), vocabSize))
+	return result
+}
+
 type ModelParams struct {
 	NumGpuLayers int
 	MainGpu      int