Spaces:
Running
Running
| { | |
| "built": "2026-11-07", | |
| "repos": 50, | |
| "groups": [ | |
| { | |
| "slug": "asr", | |
| "name": "ASR models and fine-tunes", | |
| "tab": "ASR", | |
| "models": [ | |
| { | |
| "id": "danielrosehill/Fritter", | |
| "name": "Fritter", | |
| "type": "ASR 路 Parakeet", | |
| "summary": "English dictation tuned for technical vocabulary, including AI: model and product names, CLI and cloud terms, brands, acronyms and numbers in digits.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "nvidia/parakeet-tdt-0.6b-v2", | |
| "modified": "2026-11-05", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Fritter-Nemotron-0.6B", | |
| "name": "Fritter Nemotron 0.6B", | |
| "type": "ASR 路 Nemotron streaming", | |
| "summary": "Streaming version of Fritter, the English dictation fine-tune for technical vocabulary.", | |
| "variants": [ | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "danielrosehill/Fritter-Nemotron-0.6B-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "nvidia/nemotron-speech-streaming-en-0.6b", | |
| "modified": "2026-11-06" | |
| }, | |
| { | |
| "id": "Hebrish/Jachnun", | |
| "name": "Jachnun", | |
| "type": "ASR 路 Parakeet", | |
| "summary": "The code-switching line: English speech with romanized Hebrew words dropped in, as English speakers in Israel talk.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Jachnun-GGUF" | |
| }, | |
| { | |
| "label": "MLX", | |
| "id": "Hebrish/Jachnun-MLX" | |
| }, | |
| { | |
| "label": "ONNX", | |
| "id": "Hebrish/Jachnun-ONNX" | |
| }, | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "Hebrish/Jachnun-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-4.0", | |
| "base_model": "nvidia/parakeet-tdt-0.6b-v2", | |
| "modified": "2026-11-04" | |
| }, | |
| { | |
| "id": "Hebrish/Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-Codeswitching", | |
| "name": "Nemotron Speech Streaming 0.6B English-Hebrew Codeswitching", | |
| "type": "ASR 路 Nemotron streaming", | |
| "summary": "Streaming fine-tune for English speech with Hebrew words in it, written in canonical romanization.", | |
| "status": "Experimental", | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "other", | |
| "base_model": "nvidia/nemotron-speech-streaming-en-0.6b", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Sabich", | |
| "name": "Sabich", | |
| "type": "ASR 路 Parakeet", | |
| "summary": "The kitchen-sink line, trained on everything: English with Hebrew words in it, tech vocabulary, brand names, numbers in digits and Irish-English phrasing.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Sabich-GGUF" | |
| }, | |
| { | |
| "label": "MLX", | |
| "id": "Hebrish/Sabich-MLX" | |
| }, | |
| { | |
| "label": "ONNX", | |
| "id": "Hebrish/Sabich-ONNX" | |
| }, | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "Hebrish/Sabich-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "nvidia/parakeet-tdt-0.6b-v2", | |
| "modified": "2026-11-02" | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-asr4all-M", | |
| "name": "Sabich asr4all M", | |
| "type": "ASR 路 asr4all streaming", | |
| "summary": "Sabich on a tiny streaming base, FUTO asr4all-m.", | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "other", | |
| "base_model": "futo-org/asr4all-m", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-asr4all-S", | |
| "name": "Sabich asr4all S", | |
| "type": "ASR 路 asr4all streaming", | |
| "summary": "Sabich on a tiny streaming base, FUTO asr4all-s.", | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "other", | |
| "base_model": "futo-org/asr4all-s", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-Canary-Qwen-2.5B", | |
| "name": "Sabich Canary Qwen 2.5B", | |
| "type": "ASR 路 Canary-Qwen", | |
| "summary": "The Sabich training data applied to a larger base model, NVIDIA Canary-Qwen 2.5B.", | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-4.0", | |
| "base_model": "nvidia/canary-qwen-2.5b", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-Parakeet-110M", | |
| "name": "Sabich Parakeet 110M", | |
| "type": "ASR 路 Parakeet", | |
| "summary": "The small Sabich, on Parakeet TDT-CTC 110M.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Sabich-Parakeet-110M-GGUF" | |
| }, | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "Hebrish/Sabich-Parakeet-110M-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-4.0", | |
| "base_model": "nvidia/parakeet-tdt_ctc-110m", | |
| "modified": "2026-10-30" | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-Qwen3-ASR-1.7B", | |
| "name": "Sabich Qwen3 ASR 1.7B", | |
| "type": "ASR 路 Qwen3-ASR", | |
| "summary": "The Sabich training data applied to a larger base model, Qwen3-ASR 1.7B.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Sabich-Qwen3-ASR-1.7B-GGUF" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "apache-2.0", | |
| "base_model": "Qwen/Qwen3-ASR-1.7B", | |
| "modified": "2026-10-30" | |
| }, | |
| { | |
| "id": "Hebrish/Sabich-Rolling", | |
| "name": "Sabich Rolling", | |
| "type": "ASR 路 Parakeet", | |
| "summary": "Sabich trained forward: each release continues from the previous weights on all the recordings collected so far.", | |
| "variants": [ | |
| { | |
| "label": "ONNX", | |
| "id": "Hebrish/Sabich-Rolling-ONNX" | |
| }, | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "Hebrish/Sabich-Rolling-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "Hebrish/Sabich", | |
| "modified": "2026-11-05" | |
| }, | |
| { | |
| "id": "Hebrish/Shakshuka", | |
| "name": "Shakshuka", | |
| "type": "ASR 路 Nemotron streaming", | |
| "summary": "The streaming line: text appears while you speak. Trained on the same everything-in-it data as Sabich.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Shakshuka-GGUF" | |
| }, | |
| { | |
| "label": "sherpa-onnx int8", | |
| "id": "Hebrish/Shakshuka-sherpa-onnx-int8" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "nvidia/nemotron-speech-streaming-en-0.6b", | |
| "modified": "2026-11-04" | |
| }, | |
| { | |
| "id": "Hebrish/Whisper-Hebrish-Small-ACFT", | |
| "name": "Whisper Hebrish Small ACFT", | |
| "type": "ASR 路 Whisper", | |
| "summary": "Phone-sized Whisper fine-tune for English speech with Hebrew words, built to run inside FUTO Keyboard and FUTO Voice Input on Android.", | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "mit", | |
| "base_model": "openai/whisper-small.en", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| } | |
| ] | |
| }, | |
| { | |
| "slug": "small", | |
| "name": "Small models and classifiers", | |
| "tab": "Small models", | |
| "models": [ | |
| { | |
| "id": "danielrosehill/Gemma-3-270M-ASR-Paragraph-Breaks", | |
| "name": "Gemma 3 270M ASR Paragraph Breaks", | |
| "type": "Text clean-up", | |
| "summary": "Adds paragraph breaks to flat speech-to-text output and changes nothing else.", | |
| "variants": [ | |
| { | |
| "label": "ONNX", | |
| "id": "danielrosehill/Gemma-3-270M-ASR-Paragraph-Breaks-ONNX" | |
| } | |
| ], | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "gemma", | |
| "base_model": "google/gemma-3-270m-it", | |
| "modified": "2026-10-21" | |
| }, | |
| { | |
| "id": "danielrosehill/Gemma-3-270M-Email-Key-Line-Bolding", | |
| "name": "Gemma 3 270M Email Key Line Bolding", | |
| "type": "Text clean-up", | |
| "summary": "Puts the key parts of a dictated email in Markdown bold, and changes nothing else.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "gemma", | |
| "base_model": "google/gemma-3-270m-it", | |
| "modified": "2026-10-21", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Hebrew-Latin-Token-Classifier", | |
| "name": "Hebrew-Latin Token Classifier", | |
| "type": "Token classifier", | |
| "summary": "Finds Hebrew words written in Latin letters inside English text.", | |
| "owner": "Hebrish", | |
| "gated": false, | |
| "license": "mit", | |
| "base_model": "FacebookAI/xlm-roberta-base", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Hey-Yossi-microWakeWord", | |
| "name": "Hey Yossi microWakeWord", | |
| "type": "Wake word", | |
| "summary": "Custom \"Hey Yossi\" wake word for microWakeWord: a streaming, int8 TFLite model of 61 KB.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": null, | |
| "modified": "2026-10-29", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Hey-Yossi-openWakeWord", | |
| "name": "Hey Yossi openWakeWord", | |
| "type": "Wake word", | |
| "summary": "Custom \"Hey Yossi\" wake word for openWakeWord, a small classifier that runs in real time on any CPU.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": null, | |
| "modified": "2026-10-29", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/inventory-spec-extractor", | |
| "name": "Inventory Spec Extractor", | |
| "type": "Token classifier", | |
| "summary": "Finds structured attributes in one line of a free-text product specification. 33M parameters, int8 ONNX, about 34 MB.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "apache-2.0", | |
| "base_model": "BAAI/bge-small-en-v1.5", | |
| "modified": "2026-10-15", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/MiniLM-L12-ASR-Paragraph-Breaks", | |
| "name": "MiniLM L12 ASR Paragraph Breaks", | |
| "type": "Token classifier", | |
| "summary": "Decides, word by word, where a new paragraph starts in flat speech-to-text output. A classifier, not a generator.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "mit", | |
| "base_model": "microsoft/MiniLM-L12-H384-uncased", | |
| "modified": "2026-10-24", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/S1-ASR-Cleanup", | |
| "name": "S1 ASR Cleanup", | |
| "type": "Text clean-up", | |
| "summary": "Fine-tune of Superwhisper's S1-mini (Qwen3-0.6B) that cleans raw speech-to-text dictation the way one user edits his own.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "other", | |
| "base_model": "superwhisper/s1-mini", | |
| "modified": "2026-10-30", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Voice-Note-Intent-Classifier", | |
| "name": "Voice Note Intent Classifier", | |
| "type": "Text classifier", | |
| "summary": "Multi-label classifier that reads a voice-note transcript and predicts its intent, such as shopping list, to-do or email draft, so an agent can route it.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "mit", | |
| "base_model": "BAAI/bge-base-en-v1.5", | |
| "modified": "2026-10-15", | |
| "variants": [] | |
| } | |
| ] | |
| }, | |
| { | |
| "slug": "niche", | |
| "name": "Niche and specialist models", | |
| "tab": "Specialist", | |
| "models": [ | |
| { | |
| "id": "danielrosehill/Agent-Task-Break-Detector-ModernBERT", | |
| "name": "Agent Task Break Detector (ModernBERT)", | |
| "type": "Agent harness", | |
| "summary": "Runs on each user turn of an AI agent session and judges whether the message is a new task, so a harness hook can start a fresh session.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "apache-2.0", | |
| "base_model": "answerdotai/ModernBERT-base", | |
| "modified": "2026-10-25", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Contextual-Sentence-Creator", | |
| "name": "Contextual Sentence Creator", | |
| "type": "Sentence generation", | |
| "summary": "Given a term, a one-line context and a count, returns that many natural English sentences using the term, as JSON.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "danielrosehill/Contextual-Sentence-Creator-GGUF" | |
| } | |
| ], | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "apache-2.0", | |
| "base_model": "ibm-granite/granite-4.2-3b", | |
| "modified": "2026-11-03" | |
| }, | |
| { | |
| "id": "danielrosehill/Dinosaur-Sloth", | |
| "name": "Dinosaur Sloth", | |
| "type": "Image LoRA 路 FLUX", | |
| "summary": "FLUX LoRA made from photos of a sloth plushie wearing a dinosaur costume.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "mit", | |
| "base_model": "black-forest-labs/FLUX.1-dev", | |
| "modified": "2025-11-06", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Herman-Poppleberry", | |
| "name": "Herman Poppleberry", | |
| "type": "Image LoRA 路 FLUX", | |
| "summary": "FLUX character LoRA made from photos of a handcrafted donkey plushie.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": null, | |
| "base_model": "black-forest-labs/FLUX.1-dev", | |
| "modified": "2025-11-06", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Jerusalem-Images", | |
| "name": "Jerusalem Images", | |
| "type": "Image LoRA 路 FLUX", | |
| "summary": "FLUX LoRA for realistic images of Jerusalem, trained on West Jerusalem streetscapes photographed in summer 2025.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "mit", | |
| "base_model": "black-forest-labs/FLUX.1-dev", | |
| "modified": "2025-11-06", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/jerusalem-poster-detector", | |
| "name": "Jerusalem Poster Detector", | |
| "type": "Object detection", | |
| "summary": "YOLO11n fine-tune that locates specific poster designs in street photographs and names the design it found.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "cc-by-4.0", | |
| "base_model": null, | |
| "modified": "2026-09-01", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "Hebrish/Meorav", | |
| "name": "Meorav", | |
| "type": "Text to speech", | |
| "summary": "Single-speaker Qwen3-TTS fine-tune on Daniel Rosehill's voice: Irish-accented English with Hebrew words mixed in.", | |
| "variants": [ | |
| { | |
| "label": "GGUF", | |
| "id": "Hebrish/Meorav-GGUF" | |
| } | |
| ], | |
| "owner": "Hebrish", | |
| "gated": true, | |
| "license": "cc-by-nc-4.0", | |
| "base_model": "Qwen/Qwen3-TTS-12Hz-1.7B-Base", | |
| "modified": "2026-10-22" | |
| }, | |
| { | |
| "id": "danielrosehill/Proximate-Pseudonymizer-0.6B", | |
| "name": "Proximate Pseudonymizer 0.6B", | |
| "type": "Privacy", | |
| "summary": "Rewrites a text with its identifying details swapped for close, plausible substitutes. Still leaves about 9% unchanged, so check every output.", | |
| "status": "Proof of concept", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": "apache-2.0", | |
| "base_model": "Qwen/Qwen3-0.6B", | |
| "modified": "2026-10-21", | |
| "variants": [] | |
| }, | |
| { | |
| "id": "danielrosehill/Tel-Aviv-Street-Style", | |
| "name": "Tel Aviv Street Style", | |
| "type": "Image LoRA 路 FLUX", | |
| "summary": "FLUX LoRA for Tel Aviv street scenes. Trigger word: TLVSCENE.", | |
| "owner": "danielrosehill", | |
| "gated": false, | |
| "license": null, | |
| "base_model": "black-forest-labs/FLUX.1-dev", | |
| "modified": "2025-11-06", | |
| "variants": [] | |
| } | |
| ] | |
| } | |
| ] | |
| } | |