mirror of
https://github.com/songquanpeng/one-api.git
synced 2025-10-15 15:30:26 +00:00
feat: add Max Tokens and Context Window Setting Options for Ollama Channel (#1694)
* Update main.go with max_tokens param * Update model.go with max_tokens param * Update model.go * Update main.go * Update main.go * Adds num_ctx param for Ollama Channel * Added num_ctx param for ollama adapter * Added num_ctx param for ollama adapter * Improved data process logic
This commit is contained in:
@@ -31,6 +31,8 @@ func ConvertRequest(request model.GeneralOpenAIRequest) *ChatRequest {
|
||||
TopP: request.TopP,
|
||||
FrequencyPenalty: request.FrequencyPenalty,
|
||||
PresencePenalty: request.PresencePenalty,
|
||||
NumPredict: request.MaxTokens,
|
||||
NumCtx: request.NumCtx,
|
||||
},
|
||||
Stream: request.Stream,
|
||||
}
|
||||
@@ -118,8 +120,10 @@ func StreamHandler(c *gin.Context, resp *http.Response) (*model.ErrorWithStatusC
|
||||
common.SetEventStreamHeaders(c)
|
||||
|
||||
for scanner.Scan() {
|
||||
data := strings.TrimPrefix(scanner.Text(), "}")
|
||||
data = data + "}"
|
||||
data := scanner.Text()
|
||||
if strings.HasPrefix(data, "}") {
|
||||
data = strings.TrimPrefix(data, "}") + "}"
|
||||
}
|
||||
|
||||
var ollamaResponse ChatResponse
|
||||
err := json.Unmarshal([]byte(data), &ollamaResponse)
|
||||
|
@@ -7,6 +7,8 @@ type Options struct {
|
||||
TopP float64 `json:"top_p,omitempty"`
|
||||
FrequencyPenalty float64 `json:"frequency_penalty,omitempty"`
|
||||
PresencePenalty float64 `json:"presence_penalty,omitempty"`
|
||||
NumPredict int `json:"num_predict,omitempty"`
|
||||
NumCtx int `json:"num_ctx,omitempty"`
|
||||
}
|
||||
|
||||
type Message struct {
|
||||
|
Reference in New Issue
Block a user