mirror of
https://github.com/songquanpeng/one-api.git
synced 2025-11-10 02:23:43 +08:00
feat: support gemini-2.0-flash
- Enhance model support by adding new entries and refining checks for system instruction compatibility. - Update logging display behavior and adjust default quotas for better user experience. - Revamp pricing structures in the billing system to reflect current model values and deprecate outdated entries. - Streamline code by replacing hardcoded values with configurations for maintainability.
This commit is contained in:
@@ -5,8 +5,10 @@ import (
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"strings"
|
||||
|
||||
"github.com/gin-gonic/gin"
|
||||
"github.com/songquanpeng/one-api/common/config"
|
||||
"github.com/songquanpeng/one-api/common/helper"
|
||||
channelhelper "github.com/songquanpeng/one-api/relay/adaptor"
|
||||
"github.com/songquanpeng/one-api/relay/adaptor/openai"
|
||||
@@ -19,15 +21,11 @@ type Adaptor struct {
|
||||
}
|
||||
|
||||
func (a *Adaptor) Init(meta *meta.Meta) {
|
||||
|
||||
}
|
||||
|
||||
func (a *Adaptor) GetRequestURL(meta *meta.Meta) (string, error) {
|
||||
var defaultVersion string
|
||||
switch meta.ActualModelName {
|
||||
case "gemini-2.0-flash-exp",
|
||||
"gemini-2.0-flash-thinking-exp",
|
||||
"gemini-2.0-flash-thinking-exp-01-21":
|
||||
defaultVersion := config.GeminiVersion
|
||||
if strings.Contains(meta.ActualModelName, "-2.0") {
|
||||
defaultVersion = "v1beta"
|
||||
}
|
||||
|
||||
|
||||
@@ -4,8 +4,38 @@ package gemini
|
||||
|
||||
var ModelList = []string{
|
||||
"gemini-pro", "gemini-1.0-pro",
|
||||
"gemini-1.5-flash", "gemini-1.5-pro",
|
||||
// "gemma-2-2b-it", "gemma-2-9b-it", "gemma-2-27b-it",
|
||||
"gemini-1.5-flash", "gemini-1.5-flash-8b",
|
||||
"gemini-1.5-pro", "gemini-1.5-pro-experimental",
|
||||
"text-embedding-004", "aqa",
|
||||
"gemini-2.0-flash-exp",
|
||||
"gemini-2.0-flash-thinking-exp", "gemini-2.0-flash-thinking-exp-01-21",
|
||||
"gemini-2.0-flash", "gemini-2.0-flash-exp",
|
||||
"gemini-2.0-flash-lite-preview-02-05",
|
||||
"gemini-2.0-flash-thinking-exp-01-21",
|
||||
"gemini-2.0-pro-exp-02-05",
|
||||
}
|
||||
|
||||
// ModelsSupportSystemInstruction is the list of models that support system instruction.
|
||||
//
|
||||
// https://cloud.google.com/vertex-ai/generative-ai/docs/learn/prompts/system-instructions
|
||||
var ModelsSupportSystemInstruction = []string{
|
||||
// "gemini-1.0-pro-002",
|
||||
// "gemini-1.5-flash", "gemini-1.5-flash-001", "gemini-1.5-flash-002",
|
||||
// "gemini-1.5-flash-8b",
|
||||
// "gemini-1.5-pro", "gemini-1.5-pro-001", "gemini-1.5-pro-002",
|
||||
// "gemini-1.5-pro-experimental",
|
||||
"gemini-2.0-flash", "gemini-2.0-flash-exp",
|
||||
"gemini-2.0-flash-thinking-exp-01-21",
|
||||
}
|
||||
|
||||
// IsModelSupportSystemInstruction check if the model support system instruction.
|
||||
//
|
||||
// Because the main version of Go is 1.20, slice.Contains cannot be used
|
||||
func IsModelSupportSystemInstruction(model string) bool {
|
||||
for _, m := range ModelsSupportSystemInstruction {
|
||||
if m == model {
|
||||
return true
|
||||
}
|
||||
}
|
||||
|
||||
return false
|
||||
}
|
||||
|
||||
@@ -132,9 +132,16 @@ func ConvertRequest(textRequest model.GeneralOpenAIRequest) *ChatRequest {
|
||||
}
|
||||
// Converting system prompt to prompt from user for the same reason
|
||||
if content.Role == "system" {
|
||||
content.Role = "user"
|
||||
shouldAddDummyModelMessage = true
|
||||
if IsModelSupportSystemInstruction(textRequest.Model) {
|
||||
geminiRequest.SystemInstruction = &content
|
||||
geminiRequest.SystemInstruction.Role = ""
|
||||
continue
|
||||
} else {
|
||||
content.Role = "user"
|
||||
}
|
||||
}
|
||||
|
||||
geminiRequest.Contents = append(geminiRequest.Contents, content)
|
||||
|
||||
// If a system message is the last message, we need to add a dummy model message to make gemini happy
|
||||
|
||||
@@ -1,10 +1,11 @@
|
||||
package gemini
|
||||
|
||||
type ChatRequest struct {
|
||||
Contents []ChatContent `json:"contents"`
|
||||
SafetySettings []ChatSafetySettings `json:"safety_settings,omitempty"`
|
||||
GenerationConfig ChatGenerationConfig `json:"generation_config,omitempty"`
|
||||
Tools []ChatTools `json:"tools,omitempty"`
|
||||
Contents []ChatContent `json:"contents"`
|
||||
SafetySettings []ChatSafetySettings `json:"safety_settings,omitempty"`
|
||||
GenerationConfig ChatGenerationConfig `json:"generation_config,omitempty"`
|
||||
Tools []ChatTools `json:"tools,omitempty"`
|
||||
SystemInstruction *ChatContent `json:"system_instruction,omitempty"`
|
||||
}
|
||||
|
||||
type EmbeddingRequest struct {
|
||||
|
||||
Reference in New Issue
Block a user