USC-PSI-Lab/humanoid-everyday
Preview • Updated • 16k • 51
Daniel sattarnejad
Daniel-Research
AI & ML interests
I am interested in operationalizing concepts from economic sociology — specifically situational analysis and mapping — into AI models by developing annotated datasets and structured frameworks (actors, discourses, institutional structures, silences) that can be used for fine-tuning and domain adaptation.”
Recent Activity
liked a dataset about 12 hours ago
USC-PSI-Lab/humanoid-everyday liked a dataset 5 days ago
hamzabagirsakci/turkish-court-decisions liked a dataset 8 days ago
nvidia/Nemotron-RL-Agentic-Terminal-Pivot-v1Organizations
None yet
liked a
dataset about 12 hours ago
liked a
dataset 5 days ago
liked a
dataset 8 days ago
liked a
dataset 9 days ago
liked a
model 20 days ago
liked 4
datasets 28 days ago
liked 4
datasets about 1 month ago
reacted to salma-remyx's post with ❤️ about 1 month ago
Post
530
Agents are helping me scale the search for my next great idea.
Outrider is a GitHub Action for recommending methods and drafting PRs for your target repo.
VQASynth is an open-source multimodal dataset synthesis pipeline, reflecting the design detailed in SpatialVLM: https://spatial-vlm.github.io/
Outrider helped me discover a pipeline simplification using the recent work: InstructSAM
Check out their repo: https://github.com/DCDmllm/InstructSAM
Now I can focus on the remaining challenges, like porting the new architecture with .gguf conversion
https://github.com/smellslikeml/sam3cpp/blob/feat/instructsam-decoder/docker/Dockerfile
Outrider is a GitHub Action for recommending methods and drafting PRs for your target repo.
VQASynth is an open-source multimodal dataset synthesis pipeline, reflecting the design detailed in SpatialVLM: https://spatial-vlm.github.io/
Outrider helped me discover a pipeline simplification using the recent work: InstructSAM
Check out their repo: https://github.com/DCDmllm/InstructSAM
Now I can focus on the remaining challenges, like porting the new architecture with .gguf conversion
https://github.com/smellslikeml/sam3cpp/blob/feat/instructsam-decoder/docker/Dockerfile
- 1 reply
reacted to kanaria007's post with ❤️ about 1 month ago
Post
128
✅ Article highlight: *Honest Downgrades: How to Say Less Without Lying* (art-60-259, v0.1)
TL;DR:
This article argues that a weaker claim can be the more trustworthy result.
When evidence, scope, freshness, or boundary visibility cannot support the requested wording, a governed system should not overclaim—and it should not always collapse into silence. 259 treats downgrade as claim-sizing: preserve the strongest statement that remains useful, bounded, and true.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
• prevents weak evidence from being inflated into strong assurance
• distinguishes honest narrowing from rejection or refusal
• preserves useful truth when the strongest claim fails
• makes freshness, scope, and boundary limits visible
• aligns runtime, review, assurance, and publication language
What’s inside:
• the distinction between DEGRADE, REJECT, NOT ADMISSIBLE, and OUT OF SCOPE
• a practical ladder for finding a supportable lower claim
• downgrade causes: evidence insufficiency, scope narrowing, freshness decay, and boundary uncertainty
• claim, artifact, and review-posture downgrades
• limit-disclosure notes and claim-downgrade registers
• wording patterns for saying less without becoming vague
Key idea:
Do not say:
*“we cannot prove the strongest version, so we will either publish it anyway or say nothing.”*
Say:
*“the requested claim is too strong for the available support, but this narrower claim remains admissible under these scope, freshness, evidence, and reliance limits.”*
Honest systems do not always speak strongly.
They speak no larger than their support.
TL;DR:
This article argues that a weaker claim can be the more trustworthy result.
When evidence, scope, freshness, or boundary visibility cannot support the requested wording, a governed system should not overclaim—and it should not always collapse into silence. 259 treats downgrade as claim-sizing: preserve the strongest statement that remains useful, bounded, and true.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
• prevents weak evidence from being inflated into strong assurance
• distinguishes honest narrowing from rejection or refusal
• preserves useful truth when the strongest claim fails
• makes freshness, scope, and boundary limits visible
• aligns runtime, review, assurance, and publication language
What’s inside:
• the distinction between DEGRADE, REJECT, NOT ADMISSIBLE, and OUT OF SCOPE
• a practical ladder for finding a supportable lower claim
• downgrade causes: evidence insufficiency, scope narrowing, freshness decay, and boundary uncertainty
• claim, artifact, and review-posture downgrades
• limit-disclosure notes and claim-downgrade registers
• wording patterns for saying less without becoming vague
Key idea:
Do not say:
*“we cannot prove the strongest version, so we will either publish it anyway or say nothing.”*
Say:
*“the requested claim is too strong for the available support, but this narrower claim remains admissible under these scope, freshness, evidence, and reliance limits.”*
Honest systems do not always speak strongly.
They speak no larger than their support.
liked a
dataset about 1 month ago
liked 2
models about 2 months ago
liked 2
datasets 2 months ago