AI & ML interests

Small language models, code generation, Rust, fine-tuning, reinforcement learning, benchmarks

Recent Activity

cMajors  published a model 4 days ago
maylore-ai/Temper-2-Sapient-GGUF
cMajors  published a model 4 days ago
maylore-ai/Temper-2-Sapient
cMajors  updated a model 4 days ago
maylore-ai/Temper-2-Sapient
View all activity

Organization Card

Maylore

Maylore develops compact language models specialized in narrow tasks that run on most consumer PCs and laptops. The Temper models write Rust and run locally in LM Studio, Ollama or llama.cpp.

Latest releases

  • Temper-2-Sapient 7B model fine-tuned from Qwen2.5-Coder-7B-Instruct. Among the open models up to 14B parameters it is the strongest on HumanEval-RS, MBPP-RS and Temper Bench v2, and on MBPP-RS it keeps pace with Gemma 4 26B-A4B.
  • Temper-2-Sapient-GGUF GGUF Q8_0 and BF16 for local use.
  • Temper-2-Nodel 4B model fine-tuned from Qwen3.5-4B. Solves 80% more Temper Bench v2 tasks than the original Qwen3.5-4B and beats Qwen3.5-9B on it, a model more than twice its size.
  • Temper-2-Nodel-GGUF GGUF Q8_0 and BF16 for local use.

Website: maylore.com

datasets 0

None public yet