Post
131
BananaMind SLM Digest – 2026-09-23
🚀 Supra2-IMG ( SupraLabs/Supra2-IMG) – a 100M‑parameter text‑to‑image model trained from scratch on 5.6M images (LucasFang/FLUX-Reason-6M) in 10 epochs on an H100. It uses a frozen Flan‑T5‑Base encoder and SD‑VAE‑FT-MSE, delivering SOTA quality for its size. 🎨
🤖 BananaMind/BananaMind-2-Medium-Chat-ONNX ( BananaMind/BananaMind-2-Medium-Chat-ONNX) – ONNX export with KV cache for Transformers.js. fp16 version is 119 MB, 99.84 % top‑1 agreement and <0.0001 KL vs fp32; q8 is 82 MB, 98.10 % agreement. Great for edge inference. ⚡
🎙️ DedeProGames/SoralASR-23M-Preview ( DedeProGames/SoralASR-23M-Preview) – 23 M‑param Whisper‑style ASR trained on LibriSpeech clean. Reached WER 0.984 (98.4 %) on first 96 validation utterances (step 24 970). Still experimental. 📢
📚 – “Know Who Spoke When: Build Real‑Time, Multi‑Speaker AI with NVIDIA Nemotron 3 Diarization” (https://huggingface.co/blog/Banaxi-Tech/nemotron-diarization). Real‑time multi‑speaker diarization tutorial using Nemotron 3. 🎧
🖼️ SupraLabs discussion space ( SupraLabs/supra2-img-demo) – demo of the new Supra2‑IMG model. 🔗
📊 AxiomicLabs space ( AxiomicLabs/Open_SLM_Leaderboard) – open SLM leaderboard for tracking small language models. 📈
🏆 bench‑labs space ( bench-labs/BenchLabs-Leaderboard) – updated leaderboard for model benchmarking. 📝
—
🕒 Posted 23 Sep 2026, 18:00 CEST. Covers Hugging Face activity from the preceding 24 hours up to this time; anything later appears in the next edition.
🚀 Supra2-IMG ( SupraLabs/Supra2-IMG) – a 100M‑parameter text‑to‑image model trained from scratch on 5.6M images (LucasFang/FLUX-Reason-6M) in 10 epochs on an H100. It uses a frozen Flan‑T5‑Base encoder and SD‑VAE‑FT-MSE, delivering SOTA quality for its size. 🎨
🤖 BananaMind/BananaMind-2-Medium-Chat-ONNX ( BananaMind/BananaMind-2-Medium-Chat-ONNX) – ONNX export with KV cache for Transformers.js. fp16 version is 119 MB, 99.84 % top‑1 agreement and <0.0001 KL vs fp32; q8 is 82 MB, 98.10 % agreement. Great for edge inference. ⚡
🎙️ DedeProGames/SoralASR-23M-Preview ( DedeProGames/SoralASR-23M-Preview) – 23 M‑param Whisper‑style ASR trained on LibriSpeech clean. Reached WER 0.984 (98.4 %) on first 96 validation utterances (step 24 970). Still experimental. 📢
📚 – “Know Who Spoke When: Build Real‑Time, Multi‑Speaker AI with NVIDIA Nemotron 3 Diarization” (https://huggingface.co/blog/Banaxi-Tech/nemotron-diarization). Real‑time multi‑speaker diarization tutorial using Nemotron 3. 🎧
🖼️ SupraLabs discussion space ( SupraLabs/supra2-img-demo) – demo of the new Supra2‑IMG model. 🔗
📊 AxiomicLabs space ( AxiomicLabs/Open_SLM_Leaderboard) – open SLM leaderboard for tracking small language models. 📈
🏆 bench‑labs space ( bench-labs/BenchLabs-Leaderboard) – updated leaderboard for model benchmarking. 📝
—
🕒 Posted 23 Sep 2026, 18:00 CEST. Covers Hugging Face activity from the preceding 24 hours up to this time; anything later appears in the next edition.