Article
Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents
MultiverseComputingCAI
• • 26matt goeckel PRO
mgoeckel
AI & ML interests
Building in a cave with a box of scraps
Recent Activity
upvoted an article about 19 hours ago
Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents liked a model 10 days ago
google/embeddinggemma-2 liked a model 15 days ago
decision-tune/decisiontune-1.0Organizations
None yet
upvoted an article about 19 hours ago
liked a
model 10 days ago
liked a
model 15 days ago
liked 8
datasets 19 days ago
choyiny/decidebench
Viewer • Updated • 697 • 1.6k • 4
nvidia/Nemotron-PII
Viewer • Updated • 200k • 5.78k • 125
jaredpalmer/kev-suites
Preview • Updated • 6.6k • 10
typesafeAI/evalsafe-customer-service
Viewer • Updated • 5.33k • 1.6k • 3
typesafeAI/evalsafe-security-incidents
Viewer • Updated • 4.22k • 1.44k • 3
typesafeAI/evalsafe-agent-trace-observability
Viewer • Updated • 2.23k • 1.42k • 3
typesafeAI/evalsafe-invoice-processing
Viewer • Updated • 8.37k • 1.77k • 7
typesafeAI/evalsafe-onet
Viewer • Updated • 75.2k • 1.1k • 8
reacted to pollix's post with 🚀 19 days ago
Post
6,453
stuntd 0.1.2 is out 🎉
stuntd sits in front of your LLM, learns its typed decisions and answers the confident ones locally with a small head on the Laya encoder by @convaiinnovations . About 20ms on GPU and 60ms on CPU, and anything it isn't sure about still goes to the big model.
New in 0.1.2:
- decisions with several fields, like category + urgency + needs_human in one call, answered locally only when every field is sure
- the Anthropic Messages API learns too, not only OpenAI
- auto_retrain: the daemon retrains a site in the background once enough new traffic comes in, so collect, train, shadow and live run on their own
- serve --lazy loads the checkpoint on the first request
Try it in the browser: pollix/stuntd
Code: https://github.com/bladedevoff/stuntd
stuntd sits in front of your LLM, learns its typed decisions and answers the confident ones locally with a small head on the Laya encoder by @convaiinnovations . About 20ms on GPU and 60ms on CPU, and anything it isn't sure about still goes to the big model.
New in 0.1.2:
- decisions with several fields, like category + urgency + needs_human in one call, answered locally only when every field is sure
- the Anthropic Messages API learns too, not only OpenAI
- auto_retrain: the daemon retrains a site in the background once enough new traffic comes in, so collect, train, shadow and live run on their own
- serve --lazy loads the checkpoint on the first request
Try it in the browser: pollix/stuntd
Code: https://github.com/bladedevoff/stuntd
pip install -U stuntd
- 1 reply
liked 2
models 19 days ago
updated a
collection 19 days ago
updated 3
models 19 days ago
liked a
model 19 days ago
