XHToken/Spark-X2.6-4B
Text Generation • 4B • Updated • 3.49k • 495
Ilides
Ilides
AI & ML interests
None yet
Recent Activity
liked a model 3 days ago
XHToken/Spark-X2.6-4B liked a model 3 days ago
OpenMOSS-Team/MOSS-Image-SFT liked a model 6 days ago
baidu/Unlimited-OCROrganizations
None yet
liked 2
models 3 days ago
liked a
model 6 days ago
updated a
model 9 days ago
published a
model 9 days ago
liked a
model 15 days ago
updated 3
models 17 days ago
published 3
models 17 days ago
liked a
dataset 18 days ago
updated a
model 19 days ago
published a
model 19 days ago
updated a
model 19 days ago
published a
model 19 days ago
reacted to comgen42's post with 🔥 19 days ago
Post
3,268
🐻 Kodiak-v0.3-1B: seven new kinds of decision, measured on real data.
Kodiak is an open 1B encoder that answers typed questions with calibrated confidence, or says "can't tell". New in v0.3:
• pairwise judge (which answer is better?)
• long-answer hallucination checks
• stance and sarcasm
• policy violation and refund-eligibility checks against your written rules
• agent step safety (run it, ask first, or never)
We stopped trusting our own synthetic tests and judged v0.3 on real labelled data it never trained on (RAGBench, MT-Bench human judgments, SemEval stance): 0.21 → 0.36, averaged over 3 training runs. Nothing else got worse, and when it says "can't tell" it's right 93% of the time (up from 88%).
Known limits are in the model card.
📦 cortex-agent-llc/kodiak-v0.3-1b
🎯 Accuracy mode: cortex-agent-llc/kodiak-v0.3-1b-accuracy
🕹️ Demo: comgen42/kodiak-demo
📝 What we built, and the kill that changed our process: https://cortexagent.com/blog/kodiak-v0-3-seven-new-kinds-of-decision-measured-on-real-data
Kodiak is an open 1B encoder that answers typed questions with calibrated confidence, or says "can't tell". New in v0.3:
• pairwise judge (which answer is better?)
• long-answer hallucination checks
• stance and sarcasm
• policy violation and refund-eligibility checks against your written rules
• agent step safety (run it, ask first, or never)
We stopped trusting our own synthetic tests and judged v0.3 on real labelled data it never trained on (RAGBench, MT-Bench human judgments, SemEval stance): 0.21 → 0.36, averaged over 3 training runs. Nothing else got worse, and when it says "can't tell" it's right 93% of the time (up from 88%).
Known limits are in the model card.
📦 cortex-agent-llc/kodiak-v0.3-1b
🎯 Accuracy mode: cortex-agent-llc/kodiak-v0.3-1b-accuracy
🕹️ Demo: comgen42/kodiak-demo
📝 What we built, and the kill that changed our process: https://cortexagent.com/blog/kodiak-v0-3-seven-new-kinds-of-decision-measured-on-real-data
- 4 replies
updated a
model 19 days ago
published a
model 19 days ago