2 3 35

Antonii

Apartman

https://t.me/apartman36

ApartmanN

AI & ML interests

Deep Learning Neural Network

Recent Activity

liked a model about 7 hours ago

microsoft/VibeVoice-1.5B

upvoted a collection about 7 hours ago

VibeVoice

reacted to eaddario's post with 🚀 about 16 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

View all activity

Organizations

liked a model about 7 hours ago

microsoft/VibeVoice-1.5B

Text-to-Speech • 3B • Updated Jan 22 • 269k • 2.36k

upvoted a collection about 7 hours ago

VibeVoice

Collection

Frontier Text-to-Speech Models https://microsoft.github.io/VibeVoice/ • 8 items • Updated Mar 2 • 243

reacted to eaddario's post with 🚀 about 16 hours ago

Post

2916

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.

eaddario/Qwen3.6-27B-GGUF
eaddario/Qwen3.6-35B-A3B-GGUF

upvoted a collection 5 days ago

talkie-13b

Collection

talkie-1930-13b is a vintage language model trained on pre-1931 English-language text. See https://github.com/talkie-lm/talkie to run talkie. • 3 items • Updated 14 days ago • 45

reacted to projectlosangeles's post with 🔥 8 days ago

Post

11596

🔥Check out first-of-its-kind SOTA Orpheus Morpheus preview!🔥

projectlosangeles/Orpheus-Morpheus

Easily generate variations or similar compositions from any MIDI!

Please ❤️if you enjoyed Orpheus Morpheus!

Sincerely,

Alex

reacted to danielhanchen's post with 🔥 11 days ago

Post

5259

Qwen3.6-27B is out now! Run it locally on 18GB RAM. 💜

Qwen3.6-27B surpasses Qwen3.5-397B-A17B on all major coding benchmarks.

GGUFs to run: unsloth/Qwen3.6-27B-GGUF
Guide + MLX: https://unsloth.ai/docs/models/qwen3.6

reacted to Ujjwal-Tyagi's post with 👍 12 days ago

Post

3923

We are hiring at Shirova AI. We need AI researchers and engineers to work in our research lab. Shirova AI is a research lab in India, so we can help our researchers move to nearby workspaces or let them work from home without ever coming to the lab. We're building our founding team, so the pay will be good. You can learn, so don't hesitate to mail us at: careers@shirova.com