udgamlabs/udgam-english-125m
Text Generation • Updated • 1
Early experiments
Our first releases, kept public for transparency, including negative results.
Note Our first English model trained from scratch. It missed its target (it did not beat SmolLM2 or Sarvam-1); weights and full results are shared openly.
udgamlabs/hinglish-commands-0.6b
Text Generation • Updated • 1Note Not trained from scratch: a LoRA adapter for Qwen3-0.6B that turns Romanized Hinglish or English requests into structured commands in 8 domains.