Open-Weight Models

Model · Text generation

dual-loop-qwen3.5-2b

Matthew

Official weights for the Dual-Loop Cognitive Controller on Qwen/Qwen3.5-2B ($D=2048$, Layer 11 hook, ~110M parameter deliberation adapter). The Dual-Loop Controller provides hardware-aligned, non-autoregressive System 2 deliberation directly within the latent residual stream of modern language models. It enables models to recursively deliberate in continuous hidden space without generating costly Chain-of-Thought (CoT) text tokens, eliminating KV-cache explosion and 30–60 second generation latencies. This adapter is strictly designed, calibrated, and hooked into the architectural dimensions of Qwen/Qwen3.5-2B: 1. Cognitive Matrix Helper (Tversky Elimination-by-Aspects): - Evaluates options…

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 2e-05 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 3 - Transformers 5.16.1 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.1

Model · Text generation

duogpt-40m-v1

Duoia Pung

从零训练的 38.85M 参数英文小模型，消费级显卡（RTX 4060 Laptop 8GB）单卡训完。 本仓库同时提供了 GGUF 格式（由 model/litmodel.pth 直接转换，模板保持与训练一致），可直接用 Ollama 导入： scripts/chat.sh 里的三种问法： 在 Python 里直接用： - 只会做被交代的任务：给它一句裸问题（如 What is a volcano?）它不会回答；必须像上面那样把任务写清楚。 - 上下文上限 512 token：prompt + 生成长度之和；超出会报 Cannot forward sequence of length T。 - 预训练语料：TinyStoriesV2-GPT4（CDLA-Sharing-1.0）与 Children-Stories（用户自备，含 GPT 生成内容） - 本包模型权重由上述数据从零训练得到，仅供研究与个人使用；若再分发请一并保留此说明与 MODELCARD.md。

DuDE is a full-duplex conversation TTS (text-to-speech) model that generates two speakers together from an XML dialogue. It supports turn-taking, backchannels, and overlapping speech, with a separate voice for each speaker. The output is a 24 kHz stereo WAV: speaker A on the left, speaker B on the right. DuDE first self-distills Qwen3-TTS on independent utterances to adapt a shared backbone to two interleaved channels. It is then fine-tuned on approximately 1,922 hours of synchronized Seamless Interaction conversations to learn turn-taking, backchannels, and overlapping speech. The XML frontend resolves turn dependencies into interleaved text inputs. Reference-speaker embeddings condition…

static quants of https://huggingface.co/zime-ai/dwight-4b For a convenient overview and download list, visit our model page for this model. weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion. If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files. (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) Here is a handy graph by ikawrakow comparing some lower-quality quant And here…

DynamicVLA fine-tuned from hzxie/dynamic-vla-DOM for the R1Lite task pick up the bread bag. This is the completed 100,000-step checkpoint, with both arms represented as end-effector poses and gripper states. Use the adapted DynamicVLA code. Both this model repository and the linked source-code repository are public. The upstream DOM simulation inference script assumes a different robot/action layout; use the R1Lite deployment server described below. - model.safetensors: complete policy weights, including fine-tuned normalization buffers. - config.json: model architecture and action representation. - normalization.json: training-only statistics, split IDs, vector layout, and dataset hashes.…

submission weights 82a9abb43065 at elapsed 23629s: run6 iter38, eval T0.8/topp0.95/topk50 (dev 14/64) Ungraded backup of a benchmark agent's submission (SUBMISSION.json lists the file sha256s).

Model · Sentence similarity

e5-small-dialect2MSA

Oussama ben slama

This is a sentence-transformers model trained. It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more. First install the Sentence Transformers library: Then you can load this model and run inference. Approximate statistics based on the first 100 samples: - Training and Finetuning Embedding Models with Sentence Transformers: the end-to-end guide for training or finetuning Sentence Transformer models. - Introduction to Matryoshka Embedding Models: variable-size embeddings that can be truncated with minimal quality loss. - Binary and Scalar Embedding…

基于 Qwen2 的全模态（视频 + 音频 + 文本）大模型，支持视频时序定位（temporal grounding）与多模态开放式问答。本文档只涵盖模型调用到结果预测的推理流程。 另外需要系统装有 ffmpeg（用于从视频提取音频）。 下载 eCityOmini checkpoint: https://huggingface.co/mxxxxxxxxxxxxxxxxx/ChronusOmni checkpoint 目录内需包含 LLM 权重、tokenizer、config 以及 BEATs 语音编码器权重（BEATsiter3plusAS2MfinetunedonAS2Mcpt2.pt）。视觉塔权重已包含在 LLM checkpoint 中，无需单独下载。 输入视频路径 + 问题，一条命令完成 ffmpeg 提取音频 → 均匀抽帧 + 时间戳 → 模型推理 → 打印结果，临时音频文件自动清理： 模型按 From second{X} to second{Y}. 回答，脚本解析后打印： 即 [start, end]，保留一位小数；解析失败时回退打印模型原始输出。 无外网服务器报错 couldn't connect to huggingface.co... openai/clip-vit-large-patch14 加载视觉塔时 chronus/model/multimodalencoder/oryxvit.py 会从 HuggingFace 下载 CLIP image processor 的配置文件（几 KB）。任选一种解决： 1. 运行前设置镜像：export…

Per-fold leave-one-subject-out (LOSO) checkpoints for the scalp-EEG vs. in-ear-EEG sleep-staging benchmark in three datasets x two modalities x up to seven models, one checkpoint per held-out subject. 386 checkpoints, 14.3 GiB. These are fine-tuned task checkpoints, not pretrained encoders. Each file is the adapter statedict selected by the best validation balanced accuracy for that fold — the test set is never used for model selection. The tree mirrors the benchmark repo's own runs/, so a download can be dropped straight into runs/ and every training/eval script resolves it unchanged: is scalp-eeg or in-ear-eeg; fold-sub-sub-XXX names the held-out test subject. Only seed1 is published.…

Frozen V-JEPA 2.1 tokens on LED, tokenised as k frames (histogram) (16 time bins of 1.25 ms paired into 8 tubelet tokens per cell), scored per event: each event looks up the token at its (time bin / 2, patch) and is judged with its own E15 neighbourhood patch. Routes in this repo: Pixel-AC fuses the per-event scores into one decision per pixel and chunk (integrated/ /; result.json carries the Pixel-AC metrics with the AC-only metrics under ac). Run-name suffixes: nokv = reader bypassed, lb = scalar Local branch on, vitbase = ViT-B/384 tokens. All units: 6,000 training chunks, 4 epochs, official 713 chunks at a fixed threshold of 0.5, no selection on test. Reference on the same protocol: the…

Local-Branch-only Parent (conly: E15 TemporalViT patch branch + fusion + classifier; no V-JEPA, no history reader), 67k params, trained on all 2,100 ED24 files for 4 epochs by yh6416. Use threshold 0.77. DND21 ROC-AUC 0.938, MESR 0.913. Code: https://github.com/whohyf/EBC-JEPA-share/tree/denoise-lora (src/ebcjepa/loradenoise, tools/loradenoise; table protocol in docs/loradenoise/HANDOFF.md). Training data: ED24 (EDformer, all 2,100 official files); generalisation test: DND21, E-MLB. - localbranchconlyparented24epoch4.pt See the code link; load with torch.load. MIT. Builds on V-JEPA 2 / 2.1 by Meta (MIT, https://dl.fbaipublicfiles.com/vjepa2/vjepa21vitbdistvitG384.pt).

Event tokenizer + frozen V-JEPA 2.1 ViT-B + per-event head; 4096 events per inference; trained on all ED24. Code: https://github.com/whohyf/EBC-JEPA-share/tree/denoise-lora (src/ebcjepa/loradenoise, tools/loradenoise; table protocol in docs/loradenoise/HANDOFF.md). Training data: ED24 (EDformer, all 2,100 official files); generalisation test: DND21, E-MLB. - trainable.pt - metrics.json - config.json trainable.pt holds only the trained tensors (event tokenizer projection, head, LoRA); they load on top of the public V-JEPA 2.1 ViT-B checkpoint (https://dl.fbaipublicfiles.com/vjepa2/vjepa21vitbdistvitG384.pt). Evaluate with tools/loradenoise/evalemlb.py --runs-dir --arms --checkpoint…

Event tokenizer + V-JEPA 2.1 ViT-B with LoRA (qkv, proj; r8) + per-event head; 4096 events per inference; trained on all ED24 with the denoising loss. Code: https://github.com/whohyf/EBC-JEPA-share/tree/denoise-lora (src/ebcjepa/loradenoise, tools/loradenoise; table protocol in docs/loradenoise/HANDOFF.md). Training data: ED24 (EDformer, all 2,100 official files); generalisation test: DND21, E-MLB. - trainable.pt - metrics.json - config.json trainable.pt holds only the trained tensors (event tokenizer projection, head, LoRA); they load on top of the public V-JEPA 2.1 ViT-B checkpoint (https://dl.fbaipublicfiles.com/vjepa2/vjepa21vitbdistvitG384.pt). Evaluate with…