fixed token count reading
This commit is contained in:
@@ -88,6 +88,11 @@ async function* completeStream(prompt, options = {}) {
|
|||||||
if (delta) yield { response: delta, done: false };
|
if (delta) yield { response: delta, done: false };
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
console.log('[llamacpp] finalTokenCount:', finalTokenCount);
|
||||||
|
yield { response: '', done: true, model: finalModel, tokenCount: finalTokenCount };
|
||||||
|
|
||||||
|
|
||||||
yield { response: '', done: true, model: finalModel, tokenCount: finalTokenCount };
|
yield { response: '', done: true, model: finalModel, tokenCount: finalTokenCount };
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|||||||
@@ -42,6 +42,7 @@ router.post('/complete/stream', async (req, res) => {
|
|||||||
// capture final metadata from the done signal
|
// capture final metadata from the done signal
|
||||||
lastModel = chunk.model ?? lastModel;
|
lastModel = chunk.model ?? lastModel;
|
||||||
tokenCount = chunk.tokenCount ?? tokenCount;
|
tokenCount = chunk.tokenCount ?? tokenCount;
|
||||||
|
console.log('[inference router] tokenCount from chunk:', chunk.tokenCount, '→', tokenCount);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|||||||
@@ -317,6 +317,7 @@ async function chatStream(externalId, userMessage, onChunk, options = {}) {
|
|||||||
console.log("[orchestration] final streamed text length:", fullText.length);
|
console.log("[orchestration] final streamed text length:", fullText.length);
|
||||||
|
|
||||||
if (fullText.trim()) {
|
if (fullText.trim()) {
|
||||||
|
console.log('[chat] tokenCount before save:', tokenCount);
|
||||||
await memory.createEpisode(session.id, userMessage, fullText, tokenCount, session.project_id ?? null);
|
await memory.createEpisode(session.id, userMessage, fullText, tokenCount, session.project_id ?? null);
|
||||||
const allEpisodes = await memory.getRecentEpisodes(session.id, 9999);
|
const allEpisodes = await memory.getRecentEpisodes(session.id, 9999);
|
||||||
triggerSummary(session, allEpisodes);
|
triggerSummary(session, allEpisodes);
|
||||||
|
|||||||
Reference in New Issue
Block a user