AI NEWS.ccoctomind-embed
Embedding model for octomind capability / skill auto-activation: `ibm-granite/granite-embedding-30m-english` (30M params, 6 layers, 384-dim, CLS-pooled, prefix-free, English) fine-tuned on trigger phrases from the octomind-tap capabilities + skills catalog and blended back into the base as a WiSE-FT model soup, which beats both the base and the raw fine-tune on the runtime gate (mean-of-top-3 cosine + threshold + margin). Training: rule-based + LLM paraphrase augmentation, one epoch of `CachedMultipleNegativesRankingLoss` (scale 10) on in-class pairs and positive-aware hard-negative triplets, `MatryoshkaLoss` over [384, 256, 192, 128, 96], then weight interpolation with the base. - `model.safetensors` + `1_Pooling/` — sentence-transformers layout (fp32). - `onnx/model.onnx` — fp32 graph. - `onnx/model_quantized.onnx` — int8 (weight-only, per-channel, plain 8-bit range); this is what the octomind runtime loads. Pool with CLS as declared in `1_Pooling/config.json`.
翻訳準備中のため原文を表示しています
Embedding model for octomind capability / skill auto-activation: `ibm-granite/granite-embedding-30m-english` (30M params, 6 layers, 384-dim, CLS-pooled, prefix-free, English) fine-tuned on trigger phrases from the octomind-tap capabilities + skills catalog and blended back into the base as a WiSE-FT model soup, which beats both the base and the raw fine-tune on the runtime gate (mean-of-top-3 cosine + threshold + margin). Training: rule-based + LLM paraphrase augmentation, one epoch of `CachedMultipleNegativesRankingLoss` (scale 10) on in-class pairs and positive-aware hard-negative triplets, `MatryoshkaLoss` over [384, 256, 192, 128, 96], then weight interpolation with the base. - `model.safetensors` + `1_Pooling/` — sentence-transformers layout (fp32). - `onnx/model.onnx` — fp32 graph. - `onnx/model_quantized.onnx` — int8 (weight-only, per-channel, plain 8-bit range); this is what the octomind runtime loads. Pool with CLS as declared in `1_Pooling/config.json`.