zack3d/big-agi
1
0
Fork 0
mirror of https://github.com/enricoros/big-AGI.git synced 2026-05-10 21:50:14 -07:00
Code Issues Packages Projects Releases Wiki Activity

Accommodate Chutes.ai / sglang parsing

Browse Source
This commit is contained in:
Enrico Ros
2025-06-02 09:32:59 -07:00
parent 604cf43627
commit edbe2e55bc
2 changed files with 10 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/modules/aix/server/dispatch/chatGenerate/parsers/openai.parser.ts
+6 -3
View File
@@ -410,7 +410,7 @@ function _fromOpenAIUsage(usage: OpenAIWire_API_Chat_Completions.Response['usage
// Input Metrics
// Input redistribution: Cache Read
if (usage.prompt_tokens_details !== undefined) {
if (usage.prompt_tokens_details) {
const TCacheRead = usage.prompt_tokens_details.cached_tokens;
if (TCacheRead !== undefined && TCacheRead > 0) {
metricsUpdate.TCacheRead = TCacheRead;
@@ -434,8 +434,11 @@ function _fromOpenAIUsage(usage: OpenAIWire_API_Chat_Completions.Response['usage
// Output Metrics
// Output breakdown: Reasoning
if (usage.completion_tokens_details?.reasoning_tokens !== undefined)
metricsUpdate.TOutR = usage.completion_tokens_details.reasoning_tokens;
if (usage.completion_tokens_details) {
const details = usage.completion_tokens_details || {};
if (details.reasoning_tokens !== undefined)
metricsUpdate.TOutR = usage.completion_tokens_details.reasoning_tokens;
}
// TODO: Output breakdown: Audio
src/modules/aix/server/dispatch/wiretypes/openai.wiretypes.ts
+4 -2
View File
@@ -394,7 +394,8 @@ export namespace OpenAIWire_API_Chat_Completions {
prompt_tokens_details: z.object({
audio_tokens: z.number().optional(),
cached_tokens: z.number().optional(),
}).optional(),
}).optional()
.nullable(), // [2025-06-02] Chutes.ai using slang server returns null for prompt_tokens_details
// [OpenAI o1, 2024-09-12] breaks down the completion tokens into components
completion_tokens_details: z.object({
@@ -403,7 +404,8 @@ export namespace OpenAIWire_API_Chat_Completions {
audio_tokens: z.number().optional(), // [OpenAI, 2024-10-01] audio tokens used in the completion (charged at a different rate)
accepted_prediction_tokens: z.number().optional(), // [OpenAI, 2024-11-05] Predicted Outputs
rejected_prediction_tokens: z.number().optional(), // [OpenAI, 2024-11-05] Predicted Outputs
}).optional(), // not present in other APIs yet
}).optional() // not present in other APIs yet
.nullable(), // [2025-06-02] no issues yet, but preventive
// [DeepSeek, 2024-08-02] context caching on disk
prompt_cache_hit_tokens: z.number().optional(),