buddhist-nlp/mitra-qwen35-embedder
Feature Extraction • 8B • Updated • 213 • 1
Mitra Qwen3.5 (2026)
Dharmamitra's Qwen3.5-9B family for classical Buddhist literature: CPT base, instruction SFT, and retrieval embedder. Trained in July 2026.
Note Retrieval embedder — SOTA on cross-lingual Buddhist retrieval (asymmetric sentence→window).
buddhist-nlp/mitra-qwen35-base-stage2
Text Generation • 9B • Updated • 28Note Stage-2 instruction model — translation and research over Buddhist source texts.
buddhist-nlp/mitra-qwen35-base-stage1
Text Generation • 9B • Updated • 32Note Stage-1 base — Qwen3.5-9B-Base + ~30B tokens Buddhist CPT; starting point for finetunes.
buddhist-nlp/mitra-qwen35-2b-base-stage2
Text Generation • 2B • Updated • 46Note 2B stage-2 base — compact translation-capable model (stage2 final, 1.4 epochs).
buddhist-nlp/mitra-qwen35-2b-embedder
Feature Extraction • 2B • Updated • 24Note 2B retrieval embedder — script-native, fast alternative to the 9B embedder.
buddhist-nlp/mitra-qwen35-it
Text Generation • 9B • Updated • 34Note General-purpose instruction/chat model — Buddhism Q&A + translation assistant.