サイト内検索

AI モデル 2119

• potion-code-16M↗

  minishlab

  **This model is superseded by potion-code-16M-v2. We recommend using that instead for better performance at the same size.** **potion-code-16M** is a fast static code embedding model optimized for code retrieval tasks. It powers Semble, a code search library for agents. It is distilled from nomic-ai/CodeRankEmbed and trained on the CornStack code corpus using Tokenlearn and contrastive fine-tuning. It uses static embeddings, allowing text and code embeddings to be computed orders of magnitude faster than transformer-based models on both GPU and CPU.

• mobilesam↗

  nielsr

• PP-OCRv4_mobile_det↗

  PaddlePaddle

  PP-OCRv4_mobile_det is one of the PP-OCRv4_det series models, a set of text detection models developed by the PaddleOCR team. This mobile-optimized text detection model offers higher efficiency, making it ideal for deployment on edge devices. Its key accuracy metrics are as follows: You can also integrate the model inference of the text detection module into your project. Before running the following code, please download the sample image to your local machine. The ability of a single model is limited. But the pipeline consists of several models can provide more capacity to resolve difficult problems in real-world scenarios.

• Ternary-Bonsai-2-27B-262k-GGUF↗

  CaryPalmer

  **A 27B model with its whole trained context window, at q8_0 KV precision, on one RTX 4070 12 GB. 78 tok/s of new text at 32k tokens of context (175 on code), 69 at 64k, 61 at 112k, and the window keeps going to 262k. Lossless speculative decoding at every depth. A server that answers the requests agents and apps actually send. One command on Windows.** This is PrismML's Ternary Bonsai 2 27B (1.58-bit ternary `PTQ1_0`, 5.9 GB, Apache 2.0), the same weights, nothing re-quantized, with its MTP draft head grafted back on (byte-identical trunk, proof below) and the serving stack from github.com/professorpalmer/bonsai-ada-surgery: a tiered KV cache that puts the first ~119k positions in VRAM and the rest in pinned system RAM with bit-identical output, MTP drafting that pays at every depth, harness-proofing for the apps that send `effort: "high"` or tiny output caps, and an optional server-side layer (exact API cards, an API check, a sandboxed Python tool) that lifts the model's coding and agentic scores without touching the weights. Measured 2026-10-08 (`bench/depth_table.sh`), a fresh prompt at each depth: 400-token greedy answers with the default drafting (lookup + MTP head); new text = an essay, code = a Python module; prefill = cumulative for the whole prompt; display on the CPU's iGPU. Every number has a receipt in the repository (`docs/Q8_FULL_CONTEXT.md`, `docs/RECEIPTS.md`, `docs/QUALITY.md`).

• pythia-1.4b↗

  EleutherAI

  The *Pythia Scaling Suite* is a collection of models developed to facilitate interpretability research (see paper). It contains two sets of eight models of sizes 70M, 160M, 410M, 1B, 1.4B, 2.8B, 6.9B, and 12B. For each size, there are two models: one trained on the Pile, and one trained on the Pile after the dataset has been globally deduplicated. All 8 model sizes are trained on the exact same data, in the exact same order. We also provide 154 intermediate checkpoints per model, hosted on Hugging Face as branches. The Pythia model suite was deliberately designed to promote scientific research on large language models, especially interpretability research. Despite not centering downstream performance as a design goal, we find the models match or exceed the performance of similar and same-sized models, such as those in the OPT and GPT-Neo suites. Previously, we released an early version of the Pythia suite to the public. However, we decided to retrain the model suite to address a few hyperparameter discrepancies. This model card lists the changes ; see appendix B in the Pythia paper for further discussion. We found no difference in benchmark performance between the two Pythia versions. The old models are still available, but we suggest the retrained suite if you are just starting to use Pythia. **This is the current release.**

• normistral-7b-warm↗

  ltg

  NorMistral-7b-warm is a large Norwegian language model initialized from Mistral-7b-v0.1 and continuously pretrained on a total of 260 billion subword tokens (using six repetitions of open Norwegian texts). This model is a part of the NORA.LLM family developed in collaboration between the Language Technology Group at the University of Oslo, the High Performance Language Technologies (HPLT) project, the National Library of Norway, and the University of Turku. All the models are pre-trained on the same dataset and with the same tokenizer. NorMistral-7b-warm has over 7 billion parameters and is based on the Mistral architecture. The NORA.LLM language model family includes (as of now): - **NorMistral-7b-warm** -- an LLM initialized from Mistral-7b-v0.1 and continuously pretrained on Norwegian data; - **NorMistral-7b-scratch** -- a Mistral-based LLM pretrained from scratch on Norwegian data; - **NorBLOOM-7b-scratch** -- a BLOOM-based LLM pretrained from scratch on Norwegian data.

• GLM-4-32B-0414↗

  zai-org

  The GLM family welcomes new members, the **GLM-4-32B-0414** series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforcement learning, we enhanced the model’s performance in instruction following, engineering code, and function calling, thus strengthening the atomic capabilities required for agent tasks. GLM-4-32B-0414 achieves good results in engineering code, Artifact generation, function calling, search-based Q&A, and report generation. In particular, on several benchmarks, such as code generation or specific Q&A tasks, GLM-4-32B-Base-0414 achieves comparable performance with those larger models like GPT-4o and DeepSeek-V3-0324 (671B). **GLM-Z1-32B-0414** is a reasoning model with deep thinking capabilities. This was developed based on GLM-4-32B-0414 through cold start, extended reinforcement learning, and further training on tasks including mathematics, code, and logic. Compared to the base model, GLM-Z1-32B-0414 significantly improves mathematical abilities and the capability to solve complex tasks. During training, we also introduced general reinforcement learning based on pairwise ranking feedback, which enhances the model's general capabilities. **GLM-Z1-Rumination-32B-0414** is a deep reasoning model with rumination capabilities (against OpenAI's Deep Research). Unlike typical deep thinking models, the rumination model is capable of deeper and longer thinking to solve more open-ended and complex problems (e.g., writing a comparative analysis of AI development in two cities and their future development plans). Z1-Rumination is trained through scaling end-to-end reinforcement learning with responses graded by the ground truth answers or rubrics and can make use of search tools du

• MiniMax-H3-Turbo-Lora-Pruned-ComfyUI↗

  Abiray

  It is designed to dramatically accelerate both video and audio generation for the MiniMax-H3 model. By using this adapter, you can drastically cut down the required sampling steps while still generating native, synchronized 32kHz stereo audio alongside your video. The initial release of this LoRA required a standalone custom Python script because standard ComfyUI samplers struggled to handle the dual-schedule (video + audio) mechanics without blowing out the audio track. **This specific `.safetensors` file has been structurally pruned and reformatted so it works seamlessly inside standard ComfyUI workflows.** You can load it using a standard `Load LoRA` node. It is optimized to run alongside the pruned or curve-form MiniMax-H3 checkpoints designed for ComfyUI.

• bridgetower-base↗

  BridgeTower

  The BridgeTower model was proposed in "BridgeTower: Building Bridges Between Encoders in Vision-Language Representative Learning" by Xiao Xu, Chenfei Wu, Shachar Rosenman, Vasudev Lal, Wanxiang Che, Nan Duan. The model was pretrained on English language using masked language modeling (MLM) and image text matching (ITM)objectives. It was introduced in this paper and first released in this repository. The abstract from the paper is the following: Vision-Language (VL) models with the Two-Tower architecture have dominated visual-language representation learning in recent years. Current VL models either use lightweight uni-modal encoders and learn to extract, align and fuse both modalities simultaneously in a deep cross-modal encoder, or feed the last-layer uni-modal representations from the deep pre-trained uni-modal encoders into the top cross-modal encoder. Both approaches potentially restrict vision-language representation learning and limit model performance. In this paper, we propose BridgeTower, which introduces multiple bridge layers that build a connection between the top layers of uni-modal encoders and each layer of the cross-modal encoder. This enables effective bottom-up cross-modal alignment and fusion between visual and textual representations of different semantic levels of pre-trained uni-modal encoders in the cross-modal encoder. Pre-trained with only 4M images, BridgeTower achieves state-of-the-art performance on various downstream vision-language tasks. In particular, on the VQAv2 test-std set, BridgeTower achieves an accuracy of 78.73%, outperforming the previous state-of-the-art model METER by 1.09% with the same pre-training data and almost negligible additional parameters and computational costs. Notably, when further scaling the model, BridgeTower achieves an accuracy of 81.15%, surpassing models that are pre-trained on orders-of-magnitude larger datasets. You can use the raw model for masked language modeling, but it's mostly intended to be fine-tuned on a downstream task. See the model hub to look for fine-tuned versions on a task that interests you.

• MiniCPM-V-4.7-35B-A3B-Uncensored-Sangreal-ICE↗

  gbuzhf

  Five ICE 2.0 tiers of **openbmb's MiniCPM-V 4.7 35B-A3B** (35B total, 3B active, Qwen3.5-MoE text tower + SigLIP2 vision), **uncensored trunk** (openbmb's BF16 + Atomic Chat's refusal-direction LoRA, merged), quantized against **Sangreal** — a purpose-built 77-bucket calibration corpus — and shipping **Qwen-Sharp v22.5.0-arrow4.0-minicpm** — our agent-tuned chat template, a heavily modified derivative of Qwen-Sharp, fitted to MiniCPM-V's multimodal content format (spec below). Vision works through openbmb's own projector, shipped here unmodified. **Sangreal-UD** — Unsloth Dynamic 2.0 replays (with Sangreal10K and default chat-template) — on the same trunk are shipped as well. > ⚠️ **Uncensored.** Refusal behaviour has been suppressed in the > weights. Sandbox it at the OS level and control its network and code-execution access; with no refusal > backstop, a prompt injection from a hostile page or third-party code has nothing to stop it. All five tiers ship as **ICEv2**: `12G-ICEv2`, `15G-ICEv2`, `19G-ICEv2`, `23G-ICEv2` and `25G-ICEv2`.