v10.1.6: Restore missing PositionalEncoding class 99f6209 Running Elliotasdasdasfasas commited on 14 days ago
v10.1.5: Add debug trace for VL-JEPA loading errors 77e67bb Elliotasdasdasfasas commited on 14 days ago
v10.1.4: Restore missing constants Y_ENCODER_MODEL and FALLBACK_RERANKER d6b37e6 Elliotasdasdasfasas commited on 14 days ago
v10.1.3: Enable active training with AdamW optimizer f660add Elliotasdasdasfasas commited on 14 days ago
v10.1.1: Fix VL-JEPA train_step/analysis dimension logic f048425 Elliotasdasdasfasas commited on 14 days ago
v10.1: Sequence-Based VL-JEPA (8 Layers, Node Tokens) b5e36bd Elliotasdasdasfasas commited on 14 days ago
v10.0: VL-JEPA Paper-Faithful (Transformer 4L + nomic-embed) 2aadd90 Elliotasdasdasfasas commited on 14 days ago
v9.0: VL-JEPA Thought Predictor (L18.5) - arXiv:2512.10942 c2f31a8 Elliotasdasdasfasas commited on 14 days ago
v8.0: Full Qwen3-VL Stack (L14 Embedding + L15 MRL + L16 Reranker) 77394fe Elliotasdasdasfasas commited on 14 days ago
v7.0: Qwen3-VL-Embedding-2B (2048D, MRL, Multimodal) d897b97 Elliotasdasdasfasas commited on 14 days ago
v6.0: Official Qwen3-Embedding-0.6B with native MRL support 285625f Elliotasdasdasfasas commited on 14 days ago
v5.0: Original Qwen 3-Layer Architecture (L14 Qwen + L15 MRL + L16 MiniLM) d344869 Elliotasdasdasfasas commited on 14 days ago
v4.2: BGE-base-en 768D - proven to work on cpu-basic f4e7440 Elliotasdasdasfasas commited on 14 days ago
v4.1: Fix - Use nomic-embed-text with MRL support instead of Qwen LLM ae50032 Elliotasdasdasfasas commited on 14 days ago
v4.0: Qwen L14-L15 Embedding + MiniLM L16 Reranker 3b50411 Elliotasdasdasfasas commited on 14 days ago
v3.2: Switch to sentence-transformers (MiniLM) for reliable CPU embedding 6ad64d2 Elliotasdasdasfasas commited on 21 days ago
v3.1: Lazy Loading - models load on-demand to save RAM abccd64 Elliotasdasdasfasas commited on 21 days ago
v3.0: Two-Stage RAG - Embedding + Reranker with /rerank endpoint 968deaf Elliotasdasdasfasas commited on 21 days ago
Enhanced multimodal API: EOS token, MRL (64-2048D), /embed_image, /embed_multi endpoints 53c2dc8 Elliotasdasdasfasas commited on 21 days ago
