54 Commits

Author SHA1 Message Date
Han Xiao
d44cec6524 fix: improve json parsing resilience and disable gemini thinking
- add jsonrepair fallback for truncated LLM output
- disable gemini built-in thinking mode (thinkingBudget: 0)
- increase token limits for errorAnalyzer, queryRewriter, serpCluster
- switch production default to gemini-2.5-flash-lite
- fix normalizeHostName to handle wildcard patterns
2025-12-13 12:07:37 +01:00
Han Xiao
579fd95fff bump to 2.5 flash light 2025-10-06 20:26:17 +02:00
Han Xiao
44dae8efb3 feat: add serpCluster integration and schema 2025-06-17 17:43:31 -07:00
Han Xiao
1fef3c26d9 refactor: replace mdFixer with finalizer and reducer, add ngram script 2025-06-11 17:02:33 -07:00
Han Xiao
7965ce1167 refactor: remove searchGrounding config and related code 2025-06-11 15:35:53 -07:00
Han Xiao
3d10f25028 feat: add agentic team mapreduce pattern 2025-06-11 13:18:00 -07:00
Han Xiao
9edf122a8c refactor: logger 2025-06-10 11:48:19 -07:00
Han Xiao
10b084ce08 fix: remove model from mdFixer config in config.json 2025-06-09 18:04:11 -07:00
Han Xiao
78b2cbb2cf feat: add mdFixer tool config and update agent logic 2025-06-09 17:47:19 -07:00
Sha Zhou
f1ef6c5cd0 fix: add early API key validation 2025-05-16 14:58:17 +08:00
Sha Zhou
3d6e6f73ea remove node memory limit 2025-05-14 09:24:10 +08:00
Sha Zhou
f62bb348d4 fix the global allContext issue 2025-05-13 16:08:32 +08:00
Sha Zhou
6f41539587 assert uid for each request 2025-05-09 18:42:35 +08:00
Han Xiao
a03e20f0bf fix: increase rate limits in jinaAiMiddleware 2025-05-07 14:26:17 +02:00
Sha Zhou
9ea355a9a5
clean up prompt text (#104) 2025-05-07 10:22:40 +08:00
Han Xiao
2534439e6e fix: increase rate limit period in jinaAiMiddleware from 60 to 120 seconds 2025-05-03 07:21:51 +02:00
Sha Zhou
03a2657fb6 increase nodejs memory limit 2025-04-29 17:50:33 +08:00
Yanlong Wang
f23be85ac2
fix: expect rough context data 2025-04-29 10:36:15 +08:00
Han Xiao
4b3ee0d9cd feat: add new files and updates 2025-04-22 18:46:40 +08:00
Han Xiao
0f0070b56e fix: update patch-express and agent implementation 2025-04-22 14:10:55 +08:00
yanlong.wang
98d83e84bb
saas: new rate limit policy 2025-03-26 11:54:15 +08:00
Han Xiao
8fa9aaa151 fix: gen obj retry 2025-03-20 09:56:21 +08:00
Han Xiao
2cf9061a39 fix: gen obj retry 2025-03-20 09:51:18 +08:00
Han Xiao
d73fc84e40 fix: gen obj retry 2025-03-20 09:38:46 +08:00
Han Xiao
b1d038faa9 fix: gen obj retry 2025-03-20 09:35:30 +08:00
Han Xiao
b213ddbce7 fix: up err model 2025-03-19 15:39:01 +08:00
Han Xiao
4b2b9774b4 Revert "perf: opt reranker"
This reverts commit e5d5fa44a86b9334bf811c3fc0fe8913a48bcc0d.
2025-03-19 15:36:55 +08:00
Yanlong Wang
7b64dde592
chore: rate limit anonymous requests to 2rpm 2025-03-18 21:19:56 +08:00
Han Xiao
e5d5fa44a8 perf: opt reranker 2025-03-18 14:12:47 +08:00
Han Xiao
51e8540b21 feat: add hostnames bw filter 2025-03-18 10:46:03 +08:00
Han Xiao
b0c07162dd fix: strict evaluator 2025-03-14 11:57:02 +08:00
Han Xiao
8df4df5b0a fix: url datetime guessing 2025-03-11 11:38:26 +08:00
Han Xiao
a986828ce4 feat: add url ranking 2025-03-04 16:55:47 +08:00
Yanlong Wang
f1b9b2f55e
fix 2025-02-27 16:38:42 +08:00
Han Xiao
c02588a92c feat: optimize prompt, coding, reflect 2025-02-24 13:16:18 +08:00
Han Xiao
528a6343e2 fix: overlength gen 2025-02-22 00:28:10 +08:00
yanlong.wang
36921f444c
jina-ai: minor fix 2025-02-20 13:59:09 +08:00
yanlong.wang
4decc9a750
search: introduce serper search provider 2025-02-20 13:49:16 +08:00
yanlong.wang
13cfd57dbb
jina-ai: omit results in context 2025-02-19 18:21:12 +08:00
Yanlong Wang
b4cf88bb4a
jina-ai: fix proxy ip 2025-02-18 20:31:40 +08:00
yanlong.wang
45230b9552
jina-ai: use node 22 2025-02-18 14:35:44 +08:00
yanlong.wang
d754762f73
jina-ai: allow anonymous access with rate limit 3qpm 2025-02-17 15:25:09 +08:00
Han Xiao
f8aa2b1353 feat: add coding tools 2025-02-17 14:19:36 +08:00
Yanlong Wang
db79e40896
jina-ai: store promptContext for future use 2025-02-14 20:35:03 +08:00
Yanlong Wang
7743accfb3
jina-ai: fix promptContext check 2025-02-14 11:10:59 +08:00
yanlong.wang
507bc38546
jina-ai: fix cors 2025-02-13 18:06:58 +08:00
yanlong.wang
970167245e
jina-ai: saas features 2025-02-13 14:58:57 +08:00
yanlong.wang
fa4dccc94d
jina-ai: fix account balance constraint 2025-02-13 11:08:29 +08:00
Han Xiao
bd77535dd9
refactor: add safe obj generation (#60)
* fix: broken markdown footnote

* refactor: safe obj generation

* test: update token tracking assertions to match new implementation

Co-Authored-By: Han Xiao <han.xiao@jina.ai>

* refactor: safe obj generation

* chore: update readme

---------

Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2025-02-13 00:33:58 +08:00
Yanlong Wang
44530a4760
llm-provider: google cloud vertex 2025-02-12 18:53:07 +08:00