John Gil Cubas's picture

John Gil Cubas

JohnRoger

·

JohnGilCubas

AI & ML interests

Natural language processing, computer social sciences

Recent Activity

liked a model about 9 hours ago

Zyphra/ZAYA1-8B

reacted to eaddario's post with 👍 about 10 hours ago

Experimental global target bits‑per‑weight quantization of Qwen/Qwen3.6-27B and Qwen/Qwen3.6-35B-A3B. Unlike standard llama.cpp quantizations that rely on fixed type heuristics (e.g., Q4_K_M), the Target BPW approach optimizes per-tensor precision where it matters the most, and produces high quality models that meet a precise global file size target. Key Advantages: - VRAM Maximization: Can generate high quality models sized exactly to fit hardware constraints (e.g., fitting the model into exactly 24GB VRAM). - Data-Driven Precision: Quantization mix is determined by actual weight error sensitivity rather than hardcoded rules, often yielding better PPL/KLD size trade-offs. Full benchmarks (PPL, KLD, ARC, GPQA, MMLU, etc.) and methodology in the models' cards. https://huggingface.co/eaddario/Qwen3.6-27B-GGUF https://huggingface.co/eaddario/Qwen3.6-35B-A3B-GGUF

liked a model 1 day ago

SL-AI/GRaPE-2-Pro_GGUF

View all activity

Organizations

upvoted a collection 1 day ago

GRaPE 2

The latest set of models in the GRaPE series • 8 items • Updated 12 days ago • 3

upvoted 2 collections 8 days ago

Ling-2.6

Ling-2.6 series is designed for real-world agents that require fast responses, strong execution, and high token efficiency, with several sized SKUs. • 4 items • Updated 7 days ago • 13

MathNet

2 items • Updated 12 days ago • 3

upvoted a collection 10 days ago

Huihui4-MoE-A4B

4 items • Updated 10 days ago • 3

upvoted 3 collections 12 days ago

Nemotron Math & Reasoning

Datasets for building models that excel at math reasoning, proofs, and quantitative problem-solving. Covers SFT, RL, and pretraining data. • 21 items • Updated 13 days ago • 5

Nemotron Supervised Fine-Tuning

SFT datasets covering math, code, chat, safety, agentic, VLM, multilingual, and specialized domains. • 38 items • Updated 13 days ago • 6

SimpleSD

Simple Self-Distillation • 3 items • Updated 30 days ago • 3

upvoted 2 collections 13 days ago

DeepSeek-V4

4 items • Updated 13 days ago • 618

MiMo-V2-Flash

MiMo-V2-Flash Series • 2 items • Updated Dec 17, 2025 • 28

upvoted 9 collections 14 days ago

Teacher Logits

Logits captured from large models to act as the teacher for distillation • 3 items • Updated Dec 15, 2025 • 12

Trinity-Large-Thinking

5 items • Updated 27 days ago • 31

DARWIN-Family

비드래프트 • 44 items • Updated 5 days ago • 17

OmniCoder

2 items • Updated Mar 12 • 6

Ternary Bonsai

1.58-bit Bonsai models • 9 items • Updated 16 days ago • 83

Granite 4.1 Language Models

Efficient language models for multilingual generation, coding, RAG, and AI assistant workflows. • 6 items • Updated 8 days ago • 45

Gemopus-4

🪐 A curated collection of lightweight multimodal Gemopus-4 models designed for edge deployment. • 6 items • Updated about 5 hours ago • 14

ERNIE 4.5

collection of ERNIE 4.5 models. • 27 items • Updated Nov 11, 2025 • 188

LLaDA2.0

10 items • Updated about 23 hours ago • 44

upvoted a collection 15 days ago

Mistral Small 4

A state-of-the-art model, open-weight, with a granular Mixture-of-Experts architecture that fuses instruct, reasoning and agentic skills. • 3 items • Updated Mar 16 • 72

upvoted an article 18 days ago

Article

Argunauts Update: Learning Formal Argument Analysis with RLVF and HIRPO

Dec 2, 2025

•

2