jungletsubasa
jungletsubasa
·
AI & ML interests
None yet
Recent Activity
new activity 11 days ago
jedisct1/MiMo-V2.5-coder-Q2-v2-MTP:very very nice and smart new activity 24 days ago
AesSedai/MiMo-V2.5-GGUF:Any llama.cpp parameters to work around the looping?Organizations
None yet
reacted to FlameF0X's post with 🔥 3 days ago
very very nice and smart
2
#1 opened 11 days ago
by jungletsubasa
Any llama.cpp parameters to work around the looping?
👍 1
2
#12 opened 24 days ago
by 
tarruda
IQuest-Coder-V1 Model Family Update 40B-Thinking
1
#1938 opened 3 months ago
by jungletsubasa
reacted to SeaWolf-AI's post with 👀🤗🤝 3 months ago
Post
4326
FINAL Bench Released: The Real Bottleneck to AGI Is Self-Correction
We release FINAL Bench, the first benchmark for measuring functional metacognition in LLMs — the ability to detect and correct one's own reasoning errors. Every existing benchmark measures final-answer accuracy. None measures whether AI knows it is wrong.
Dataset: [FINAL-Bench/Metacognitive]( FINAL-Bench/Metacognitive) | 100 Tasks | 15 Domains | 8 TICOS Types | Apache 2.0
Leaderboard: FINAL-Bench/Leaderboard
Article: https://huggingface.co/blog/FINAL-Bench/metacognitive
Core Innovation
Our 5-axis rubric separates what no prior benchmark could: MA (Metacognitive Accuracy) — the ability to say "I might be wrong", and ER (Error Recovery) — the ability to actually fix it. This maps directly to the monitoring-control model of Nelson & Narens (1990) in cognitive psychology.
Three Findings Across 9 SOTA Models
We evaluated GPT-5.2, Claude Opus 4.6, Gemini 3 Pro, DeepSeek-V3.2, Kimi K2.5, and others across 100 expert-level tasks:
1. ER Dominance. 94.8% of MetaCog gain comes from Error Recovery alone. The bottleneck to AGI is not knowledge or reasoning — it is self-correction.
2. Declarative-Procedural Gap. All 9 models can verbalize uncertainty (MA = 0.694) but cannot act on it (ER = 0.302). They sound humble but fail to self-correct — the most dangerous AI safety profile.
3. Difficulty Effect. Harder tasks benefit dramatically more from metacognition (Pearson r = -0.777, p < 0.001).
Paper: FINAL Bench: Measuring Functional Metacognitive Reasoning in LLMs
FINAL Bench is the first tool to tell apart what AI truly knows from what it merely pretends to know.
We release FINAL Bench, the first benchmark for measuring functional metacognition in LLMs — the ability to detect and correct one's own reasoning errors. Every existing benchmark measures final-answer accuracy. None measures whether AI knows it is wrong.
Dataset: [FINAL-Bench/Metacognitive]( FINAL-Bench/Metacognitive) | 100 Tasks | 15 Domains | 8 TICOS Types | Apache 2.0
Leaderboard: FINAL-Bench/Leaderboard
Article: https://huggingface.co/blog/FINAL-Bench/metacognitive
Core Innovation
Our 5-axis rubric separates what no prior benchmark could: MA (Metacognitive Accuracy) — the ability to say "I might be wrong", and ER (Error Recovery) — the ability to actually fix it. This maps directly to the monitoring-control model of Nelson & Narens (1990) in cognitive psychology.
Three Findings Across 9 SOTA Models
We evaluated GPT-5.2, Claude Opus 4.6, Gemini 3 Pro, DeepSeek-V3.2, Kimi K2.5, and others across 100 expert-level tasks:
1. ER Dominance. 94.8% of MetaCog gain comes from Error Recovery alone. The bottleneck to AGI is not knowledge or reasoning — it is self-correction.
2. Declarative-Procedural Gap. All 9 models can verbalize uncertainty (MA = 0.694) but cannot act on it (ER = 0.302). They sound humble but fail to self-correct — the most dangerous AI safety profile.
3. Difficulty Effect. Harder tasks benefit dramatically more from metacognition (Pearson r = -0.777, p < 0.001).
from datasets import load_dataset
dataset = load_dataset("FINAL-Bench/Metacognitive", split="train")Paper: FINAL Bench: Measuring Functional Metacognitive Reasoning in LLMs
FINAL Bench is the first tool to tell apart what AI truly knows from what it merely pretends to know.
- 6 replies
upvoted an article 3 months ago
Article
GGML and llama.cpp join HF to ensure the long-term progress of Local AI
- +4
ggerganov, ngxson, allozaur, lysandre, victor, julien-c
• • 506 liked a
model 4 months ago
MiniCPM-SALA
2
#1836 opened 4 months ago
by jungletsubasa
Salesforce / Llama-xLAM-2-70b-fc-r
1
#1682 opened 5 months ago
by jungletsubasa
liked 2
models 5 months ago
HyperCLOVAX-SEED-Think-32B
4
#1668 opened 5 months ago
by jungletsubasa
MiroThinker-v1.5-30B
2
#1670 opened 5 months ago
by jungletsubasa
MiroThinker-v1.5-235B
2
#1669 opened 5 months ago
by jungletsubasa
reacted to daavoo's post with 🔥 6 months ago
Post
1901
2025: The Year of Agents.
2026: The Year of Local Agents?
Relying on cloud-hosted LLMs is often overkill. While frontier models still lead in complex coding, local models are now more than capable of handling many agentic workflows—with zero latency and total privacy.
To help bridge the gap between local inference and usable agents, I’m releasing agent.cpp: https://github.com/mozilla-ai/agent.cpp
It provides minimal, high-performance building blocks for agents in C++, built directly around the awesome llama.cpp ecosystem.
Stop sending your data to a remote API. Start building and running agents on your own hardware.
2026: The Year of Local Agents?
Relying on cloud-hosted LLMs is often overkill. While frontier models still lead in complex coding, local models are now more than capable of handling many agentic workflows—with zero latency and total privacy.
To help bridge the gap between local inference and usable agents, I’m releasing agent.cpp: https://github.com/mozilla-ai/agent.cpp
It provides minimal, high-performance building blocks for agents in C++, built directly around the awesome llama.cpp ecosystem.
Stop sending your data to a remote API. Start building and running agents on your own hardware.
- 1 reply
