We're sharing/showcasing best of @huggingface models. Follow to stay in loop. Promoting Open-Source models.
Joined December 2025
- Tweets31.7K
- Following32
- Followers71.8K
- Likes623
What if an LLM could retain and build on past experiences without relying entirely on external memory systems?
That's the direction MemT 1.0 is exploring.
Built on Gated DeltaNet-2, it uses token-level memory updates to retain and refine information as it processes new inputs.
Its Mixture-of-Storage mechanism is designed to scale native memory capacity, enabling the model to carry useful experience forward and apply it to future tasks.
At 128K tokens, MemT-1.0-Nano reports 1.86× faster end-to-end inference and 66× lower persistent storage compared to a full-context KV-cache baseline.
An interesting direction for AI systems that need to retain and build on experience across interactions.
Explore the models here: huggingface.co/collections/M…
Can large language models retain information from past interactions and reuse it in the next task?
Introducing MemT-1.0: Towards Memory-Native Evolution.
This time, we've built memory directly into the model architecture.
Our MemT-1.0-Nano and MemT-1.0-Flash make persistent memory part of model computation, carrying information across interactions.
Small but mighty: 164 downloads and counting. It's a great starting point for AI music projects, offering a balance of performance and accessibility.
With 67M parameters, it's efficient and runs on modest hardware. Available in safetensors and GGUF formats, and compatible with endpoints for easy deployment.
Use it to generate MIDI sequences from text prompts, create melodies, or experiment with AI composition. Perfect for musicians, hobbyists, and developers building creative tools.
Meet Melodimodell-67M, a compact AI model that turns text into music. Built on Qwen3 and trained on the Monster-MIDI dataset, it's a fun, lightweight way to explore AI music generation.
730 downloads and growing. Merged with mergekit for enhanced roleplay performance. GGUF format means easy deployment on consumer hardware. Dive into next-level character AI today.
31B parameters, based on Blazed-Forge/Gemma-4-Sphinsikus-Chronist-31B. Quantized to GGUF for efficient inference. Apache 2.0 licensed, so you can use it freely. English language focus.
Built for roleplay and storytelling. Run it locally with transformers or GGUF backends. Create rich characters, dynamic dialogues, and endless adventures. Perfect for interactive fiction and AI companions.
Meet Gemma-4-Sphinsikus-Chronist-31B-GGUF: a 31B roleplay powerhouse in GGUF format. Merged with mergekit and ready for your local setup. If you love immersive character AI, this one's for you.
With 292 downloads and a focus on uncensored creativity, this model stands out for roleplay and writing. The GGUF format ensures it runs on consumer hardware, giving you a private, unrestricted AI experience. Try it for your next creative project.
This is a 31B parameter model merged using mergekit, then converted to GGUF format for efficient inference. It's based on the Gemma architecture and tagged for transformers, making it compatible with popular tools. A solid choice for local deployment.
Use it for creative writing, immersive roleplay, or deep reasoning tasks. It handles both reasoning and non-reasoning modes, and since it's uncensored, it won't shy away from NSFW content. Build stories, characters, or explore ideas without filters.
