github/new-api
1
0
Fork 0
mirror of https://github.com/QuantumNous/new-api.git synced 2026-04-19 13:58:38 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: refactor token estimation logic

Browse Source 
- Introduced new OpenAI text models in `common/model.go`.
- Added `IsOpenAITextModel` function to check for OpenAI text models.
- Refactored token estimation methods across various channels to use estimated prompt tokens instead of direct prompt token counts.
- Updated related functions and structures to accommodate the new token estimation approach, enhancing overall token management.
This commit is contained in:
CaIon
2025-12-02 21:34:39 +08:00
parent d3ca454c3b
commit f5b409d74f
26 changed files with 389 additions and 275 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
relay/channel/claude/relay-claude.go
View File

@@ -673,7 +673,7 @@ func HandleStreamResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
func HandleStreamFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, claudeInfo *ClaudeResponseInfo, requestMode int) {
if requestMode == RequestModeCompletion {
claudeInfo.Usage = service.ResponseText2Usage(c, claudeInfo.ResponseText.String(), info.UpstreamModelName, info.PromptTokens)
claudeInfo.Usage = service.ResponseText2Usage(c, claudeInfo.ResponseText.String(), info.UpstreamModelName, info.GetEstimatePromptTokens())
} else {
if claudeInfo.Usage.PromptTokens == 0 {
//上游出错
@@ -734,10 +734,7 @@ func HandleClaudeResponseData(c *gin.Context, info *relaycommon.RelayInfo, claud
return types.WithClaudeError(*claudeError, http.StatusInternalServerError)
}
if requestMode == RequestModeCompletion {
completionTokens := service.CountTextToken(claudeResponse.Completion, info.OriginModelName)
claudeInfo.Usage.PromptTokens = info.PromptTokens
claudeInfo.Usage.CompletionTokens = completionTokens
claudeInfo.Usage.TotalTokens = info.PromptTokens + completionTokens
claudeInfo.Usage = service.ResponseText2Usage(c, claudeResponse.Completion, info.UpstreamModelName, info.GetEstimatePromptTokens())
} else {
claudeInfo.Usage.PromptTokens = claudeResponse.Usage.InputTokens
claudeInfo.Usage.CompletionTokens = claudeResponse.Usage.OutputTokens