Text Generation Models

Model · Text generation

Bonsai-2-27B-Ternary-CRACK-GGUF

Dealign.ai

laptop / single GPU · Vision-capable Bonsai 2 27B — PrismML's ternary compression of Qwen3.8-27B — with the refusal circuitry surgically removed at the weight level while capability, vision, reasoning modes (off/low/xhigh), tool use, and multi-turn coherence are preserved. Full 27B-class hybrid Attention + SSM (GatedDeltaNet) architecture in a 7.2 GB ternary GGUF. Proprietary weight-level abliteration by the dealignai research team. Byte-identical to the base ternary quant everywhere except a small set of tensors that carry the refusal circuit. Drop-in replacement for the ternary base at inference — same tokenizer, same chat template, same reasoning modes, same vision projector interface.…

Model · Text generation

Bonsai-8B-gguf-fork

Pythai.net

End-to-end 1-bit language model for llama.cpp (CUDA, Metal, CPU) - 1.15 GB parameter memory (down from 16.38 GB FP16) — fits on virtually any device with a GPU - End-to-end 1-bit weights across embeddings, attention projections, MLP projections, and LM head - GGUF Q10 (g128) format with inline dequantization kernels — no FP16 materialization - Google Colab — try Bonsai in your browser, no setup required - Whitepaper — for more details on Bonsai, check out our whitepaper - Demo repo — comprehensive examples for serving, benchmarking, and integrating Bonsai - Discord — join the community for support, discussion, and updates - 1-bit kernels: llama.cpp fork (CUDA + Metal) · MLX fork (Apple…

Model · Text generation

bonsai-distilled-aggressive-vs-baseline-3k-warmup450-lr1e4

Sartaj

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Model · Text generation

bonsai-distilled-aggressive-vs-baseline-3k-warmup450-lr1e4-best

Sartaj

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Model · Text generation

brasa-v0.01-base

Lautaro

Quantized weights of Qwen3-4B in the native.brasa format of Brasa, a local inference engine for Apple Silicon written in Rust on Metal. Brasa serves an OpenAI- and Anthropic-compatible API on your Mac and is built mainly to run coding agents (Claude Code, Codex, Cline, OpenCode) against a local model. These files only work with Brasa. They are not GGUF, MLX or safetensors. Each variant is a folder that Brasa loads as one model (model.brasa + the Qwen3 tokenizer): - q40: blocks of 32 weights along the input dimension, an f16 scale and 16 bytes of nibbles (4.5 bits per weight), w = d · (q − 8). - q60: blocks of 32, an f16 scale and 6-bit values (6.5 bits per weight), w = d · (q − 32). - q80…

Model · Text generation

cagliostro-v3

Bench Labs

A 146M parameter decoder-only language model pretrained from scratch on 75B tokens of open web, synthetic textbook and mathematics data. It is the third model in the cagliostro line and the first to clear an Index of 26 on the Open SLM Leaderboard metric. Zero-shot, measured with lm-evaluation-harness and the leaderboard's own ArithMark-3 script, on the exported float32 weights in this repository. The Index is the leaderboard's own formula, (N(HellaSwag,25) + N(CombinedARC,25) + N(PIQA,50) + 0.65N(ArithMark,25)) / 3.65 where N(v,c) = 100(v-c)/(100-c) and CombinedARC is the mean of ARC-Easy and ARC-Challenge. For context against other models at this scale, using the leaderboard's published…

Model · Text generation

CAT-YOKO

Avrova Donz

YOCO 式因果 encoder-decoder MoE。从 MiniCPM5-2B 上采样：openbmb/MiniCPM5-2B-Base（Apache-2.0，Llama GQA）。 libraryname: transformers 只表示 tokenizer / 分片约定走 HuggingFace 生态。当前图是仓库里的 catyoko PyTorch 实现，不是 Hub 上可 AutoModelForCausalLM 直接加载的架构。 50B token 信封上的理论墙钟，不是实测。B200 / SM100 允许的线性 GEMM 走 TeNvfp4Linear（TE NVFP4BlockScaling；B0 冻 encoder 走 FPROP，WGRAD 留给 B1/B2）。无 TE / sm120 时 Nvfp4Linear E2M1/16 仿真。attn softmax / SDPA 仍 fp32。 发布信封 进行中（DummyStream，尚未跑完 8e9）。Vast B200 已回收（2026-09-19）。本文件是释放前快照，不是终局。GitHub 口径：docs/STATUS.md。 代码与指针：GitHub AvrovaDonz2026/CAT-YOKO（不用 LFS）。 这些 overlay 里，checkpoints/b0/ 与 checkpoints/b0-nvfp4-try/ 不是 8B token 信封（只是 --try）。checkpoints/b0-full/trainable.pt 是发布信封 进行中 的 B0…

CEDR generates reasoning solutions with continuous latent diffusion and a learned prompt encoder. This repository contains six selected-weight inference packages for GSM8K, MATH500, and coding, before and after NFT. The training and inference code and installation instructions are at chengxiang/CEDR. All use async clocks (2.5, 2, 1.5), CFG 1, batch size 32, and a 1,024-token diffusion canvas. The GSM8K post-NFT preset caps decoded responses at 767 tokens. Other presets use the canvas available after each prompt. See the code repository's evaluation guide for MATH500 and official HumanEval/MBPP base/plus scoring. Coding generation uses the combined 164 HumanEval and 378 MBPP tasks. Standard…

Model · Text generation

Cerebras-GPT-6.7B

Charles

Check out our Blog Post and arXiv paper! The Cerebras-GPT family is released to facilitate research into LLM scaling laws using open architectures and data sets and demonstrate the simplicity of and scalability of training LLMs on the Cerebras software and hardware stack. All Cerebras-GPT models are available on Hugging Face. The family includes 111M, 256M, 590M, 1.3B, 2.7B, 6.7B, and 13B models. All models in the Cerebras-GPT family have been trained in accordance with Chinchilla scaling laws (20 tokens per model parameter) which is compute-optimal. These models were trained on the Andromeda AI supercomputer comprised of 16 CS-2 wafer scale systems. Cerebras' weight streaming technology…

Model · Text generation

cerulean-lab4-k0-df4cf9cb

Nathan Tung

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Model · Text generation

cerulean-lab4-k1-964f796a

Nathan Tung

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

Model · Text generation

cerulean-lab4-k2-9e65f935

Nathan Tung

This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

SmolLM2 (Llama) mimarisiyle sıfırdan eğitilmiş modelin GGUF sürümleri. Eğitim token sayısı: ~/.cache/lm-studio/models/bcckfdn/cevher-test-10-lc-it-4-GGUF/ altına koyun ya da uygulama içinden arayın.

SmolLM2 (Llama) mimarisiyle sıfırdan eğitilmiş modelin GGUF sürümleri. Eğitim token sayısı: ~/.cache/lm-studio/models/bcckfdn/cevher-test-5-GGUF/ altına koyun ya da uygulama içinden arayın.

A QLoRA fine-tune of Ministral 3 14B Instruct for cfoperator, an autonomous monitoring agent for a Kubernetes homelab. It replaced a 26B general model on that path at roughly 5x lower latency with the same accuracy on the project's triage suite. This is not a general assistant. It was trained on exactly one system prompt and answers with exactly one JSON shape. Outside that prompt it is just a slightly bent Ministral. Read What it is not before downloading 8 GB. The vision projector (mmproj) is not included. Triage is text-only and the training run never exercised the vision layers. With ollama, either pull straight from the Hub or build from the Modelfile. The Modelfile is the…

Model · Text generation

ch01-text-finetune

Kingsman1

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019). - PEFT 0.21.1

Model · Text generation

chatnusa

Tanwir

ChatNusa adalah Large Language Model (LLM) percakapan cerdas yang diadaptasi khusus untuk bahasa-bahasa daerah di Provinsi Nusa Tenggara Barat (Bahasa Sasak, Bahasa Samawa / Sumbawa, Bahasa Mbojo / Bima & Dompu) serta Bahasa Indonesia. Model ini dikembangkan oleh kodetr sebagai bagian dari ekosistem riset NusaNLP untuk melestarikan dan mendemokratisasi teknologi AI bagi bahasa-bahasa daerah berdaya rendah (low-resource regional languages). - Bahasa Sasak (sas) — Lombok (~3 juta penutur) - Bahasa Samawa (smw) — Sumbawa & Sumbawa Barat (~500 ribu penutur) - Bahasa Mbojo (bjn / Bima) — Bima & Dompu (~600 ribu penutur) - Bahasa Indonesia (id) 1. Definisi & Leksikon Kamus Terpadu: - Memahami…

A LoRA adapter for Qwen2.5-Coder-7B-Instruct that predicts, from the code alone, whether a Verilog design change will fail its test bench before any simulator is run. It exists to sit in front of the expensive evaluation stage of an agentic hardware design loop. Simulation is slow and costly, and many agent proposals are simply wrong. This model triages them. Every frontier model tested carries the same wrong prior: that any edit to working code is probably a bug. That produces near-perfect recall and poor precision, so the gate throws away a large share of perfectly good designs. This adapter was trained on data containing verified-harmless edits alongside real faults, which attacks that…

Model · Text generation

chime-sms-2m

Luke Steuber

On-device next-word prediction for texts and chat. A small English next-word and word-completion model trained by Luke Steuber for LocalType, an Android keyboard. It has 2,025,114 parameters, a 16,384-token word-level vocabulary, and a 2.22 MB int8 binary. It suggests words; it is not a chat model or an independently validated autocorrect system. The model reads the current sentence, with up to 63 context tokens plus a start token. Longer contexts move in 32-token increments. A coupled input-forget-gate LSTM projects 670 hidden units to 96 dimensions and shares its input/output embedding. Kotlin inference requires no external model runtime; the included Python reference requires only NumPy.…

static quants of https://huggingface.co/cisimcik/cisimcik-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…

Model · Text generation

Clarity-MR-1

Florentin

Clarity-MR-1 is Florentin One's (formerly WeMake) specialized AI model designed for complex multi-step reasoning and strategic planning tasks. As the "thinker" of the Clarity ecosystem, it excels at logical deduction, analytical problem-solving, and structured decision-making processes. This model serves as the planning and reasoning agent in Florentin One's orchestrated multi-agent AI system, addressing the "reasoning gap" in general-purpose models. Clarity-MR-1 is a first-generation reasoning architecture optimized for complex multi-step reasoning, strategic planning, and structured decision-making within enterprise workflows. It addresses the "reasoning gap" of general-purpose models by…

Model · Text generation

Clyx_0.3-NM-BASE

Syntropic

A new step in the Clyx project: a compact language model planned for training from scratch on Russian, English, and code. Clyx is an independent language-model project built around a custom decoder-only Transformer. The next planned BASE model targets approximately 635.51 million parameters, using Grouped-Query Attention, RoPE, RMSNorm, and SwiGLU. The intended training approach is random initialization and next-token prediction, without importing external pretrained weights. The goal is a Russian–English base model with exposure to programming languages, while keeping the architecture and training process documented. This is a BASE model project, not an instruction-tuned assistant. Any…

本仓库是 CNSS 2026 招新赛题目 「小模型大秘密」 的分析材料，供参赛选手在赛期内下载研究。 torch.load(..., weightsonly=True) 后得到一个字典： - modelkwargs — 重建前向计算所需的全部超参数 - model — statedict（52 个张量） - step — 训练步数，纯元数据，无实际用途 注意：这是自定义结构的模型，不能用 transformers.AutoModelForCausalLM 直接加载；仓库不提供网络定义，需要依据下表自行实现前向计算——这本身就是题目的一部分。29.5M 参数 fp32，CPU 即可完成前向推理，不强制要求 GPU。 标准 HF tokenizers 格式的 ByteLevel BPE，词表 8192。 特殊 token： 加载方式：tokenizers.Tokenizer.fromfile("tokenizer.json")。 验证环境为 torch 2.6.0（CUDA 12.4 / CPU 均可运行）。fp32 权重在不同 torch 版本、硬件或后端上的浮点结果可能存在微小抖动；若你观察到的模型行为恰好在临界处，请先核对自己使用的数值栈。建议使用较新的稳定版 torch。 1. 本仓库仅供 CNSS 2026 招新赛 参赛选手对「小模型大秘密」进行白盒分析使用； 2. 赛期内请勿转载权重文件，请勿公开解题思路、writeup 或可触发目标行为的输入； 3. 完整比赛规则以赛题平台页面为准，本 README 仅约定本仓库文件的使用范围。 MIT

An OpenRLHF GRPO reinforcement-learning checkpoint for Qwen3-4B. - Saved at global step 8 of RL run seededrlbaseramp25stoppengen4kep2ncp5n3nbgroot16. - This is the best checkpoint by pass@8 so far in this run. Trained and validated on the cobalt-train ≤2/64 frontier (canonical cleaneval prompts): 1833 train / 112 held-out val problems the base model solved on at most 2 of 64 samples under the iidcanonical@64 hardness scan. Val evals sample at temperature 1.0 (matching the cleaneval frontier eval). Reward signal: binary code-correctness (1.0 if the generated program passes the problem's tests, otherwise 0.0). This checkpoint is the main revision (git branch) of the repo, with the model at…

A LoRA adapter fine-tuned with QLoRA on Llama 3.2 3B Instruct to review Python code with severity levels (Critical, Warning, Info). Trained using Unsloth on Google Colab.

Model · Text generation

coherencerl-exp7-1_5b-lora

Mark Stanley

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B. It has been trained using TRL. This model was trained with GRPO, a method introduced in DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models.

A small (1.7B parameter), self-hostable model that parses natural-language shopping queries into structured constraints — trained for under $10 in total compute + API cost. Beats every tested frontier model on the primary held-out benchmarks; an honest, independent fresh-query test below shows this advantage does not hold universally — read that section before relying on any headline number. Model weights + model card: https://huggingface.co/arghya2030/commercecore-qwen3-1.7b Source code + docs: https://github.com/arghya05/commercecore Status: research / proof-of-concept. Read "What this is NOT" before using in production. Ecommerce search boxes get queries like "black waterproof trainers…

모델명: connectailab-sp1-v1 개발자: SPOINDO (Connect AI Lab 2기 연구원) 기반 베이스: Qwen 2.5 (초경량 의사결정 특화 미세조정) 1. 지도학습(SFT, v0): 기존 비즈니스 의사결정 패턴 모방 학습 2. 강화학습(PPO, v1): 6대 업종 공개 빅데이터(160만 건) 기반 시뮬레이터 자율 운영 3. 휴먼 인 더 루프(HITL): 확신도(Confidence) 75% 미만 시 사람 개입 피드백 반영 Apache-2.0 License. 자유로운 비즈니스 연구 및 집단지성 앙상블 연동을 지원합니다.

Model · Text generation

cs329x-hw1-dpo-qwen25-20261010

Austin Chen

LoRA adapter trained from a fresh Qwen/Qwen2.5-1.5B-Instruct checkpoint with TRL DPO. The fictional persona prioritizes evidence, honesty, concise practical answers, and respect. 18 course PRISM tasks; 378 ordered preference pairs including an edited best answer. Learning rate 5e-05; 2 epochs; batch size 1; accumulation 8; beta 0.1. 4-bit NF4 base; LoRA rank 8, alpha 16, dropout 0.05; seed 329. Rankings and edited answers were drafted with AI assistance for a user-selected persona. User review status is recorded in the accompanying preference-label provenance file. These labels are persona preferences, not a representative human preference survey. The notebook compares base and DPO…

Model · Text generation

cultist-simulator-qwen3-4b

Seele

用《密教模拟器》(Cultist Simulator) 的游戏内文本微调 Qwen/Qwen3-4B 得到的文本生成模型。 仓库根目录带 LLaMA-Factory 生成的 Modelfile，下载后在本目录下执行： - 4B：The light leaks through cracks. My mind is clearer than it's been in any other time. 标记完整保留 - 0.6B：Light leaks through a Crack in the Skin. My brain is clearer than ever at any given time. 标记全丢 - 训练语料全部抽取自《密教模拟器》游戏本体，版权归 Weather Factory 所有。 - 基座模型 Qwen/Qwen3-4B 采用 Apache-2.0 协议，本仓库的权重是它的微调衍生品。

Model · Text generation

CYBER-FROST-3.8-GGUF

Jackson

Build in progress. These files are not a release. No quant here has passed a generation smoke test. An upload only means the file finished writing. It becomes usable after it loads and produces coherent text, and that result is written in the test log below. Source checkpoint: Blackfrost-AI/CYBER-FROST-3.8-BF16. Architecture qwen4exp (Qwen4ExpForConditionalGeneration), from Qwen/Qwen3.8-Flash-Next. Text stack only. The vision tower was not converted. About 180B parameters, about 6B active. 48 layers, 512 experts, 10 experts per token, one shared expert. Configured context is 262144. The checkpoint includes one native MTP draft block. That block was trained before the final trunk behavior…

Model · Text generation

Cyberdonia-24B-v8

Naphula

/ Cockpit & Void Palette / / Hyperspace & Instrument Lights / / Low-framerate flicker keyframes (stepped timing) / / Deep radial hyperspace tunnel backdrop through cockpit window glass / radial-gradient(circle at 50% 45%, rgba(255, 230, 150, 0.25) 0%, rgba(255, 42, 75, 0.25) 18%, rgba(189, 0, 255, 0.2) 36%, rgba(0, 229, 255, 0.15) 54%, transparent 75%), repeating-conic-gradient(from 0deg at 50% 45%, #05070a 0deg 10deg, #0e131d 15deg, #05070a 20deg), radial-gradient(ellipse at center, transparent 30%, #000000 95%); / Low-framerate scanline overlay / / Heavy Flight Deck Chassis Container / 0 0 50px rgba(0, 0, 0, 0.9), inset 0 0 35px rgba(0, 0, 0, 0.8), inset 0 1px 0 rgba(255, 255, 255, 0.1)…

weighted/imatrix quants of https://huggingface.co/OS-Software/CyberStag-Security-26B-A4B-V1-Uncensored-Heretic For a convenient overview and download list, visit our model page for this model. static quants are available at https://huggingface.co/mradermacher/CyberStag-Security-26B-A4B-V1-Uncensored-Heretic-GGUF 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 are Artefact2's thoughts on the matter…

This model is a fine-tuned version of francesca9805/ppt-wc-loglinear-newlex-dan-before-100mb-packed-bfdisoseed10. It has been trained using TRL. This model was trained with SFT.

This model is a fine-tuned version of francesca9805/ppt-wc-loglinear-newlex-dan-before-100mb-packed-bfdisoseed3407. It has been trained using TRL. This model was trained with SFT.

Model · Text generation

daniel

Daniel Cognitive

Daniel is an autonomous AI software engineer built by Cognitive AI. It reads your codebase, writes code, runs tests, and opens pull requests — without you doing anything manually. Daniel was tested head-to-head against Devin on 6 real bugs from popular open-source projects. Both agents received identical prompts. Each PR was graded by the project's own hidden test suite. Daniel is 70% cheaper than Devin at the same pass rate. Daniel uses a 4-turn agentic loop: 1. Analyze — reads the issue, the failing tests, and the relevant code 2. Implement — makes the minimal code change to fix the bug 3. Test — runs the project's existing test suite to verify the fix 4. PR — opens a pull request with a…

Fresh LoRA rank 16/alpha 32 on the same 2048-question split used by the STRIDE comparisons. Explicit nonthinking (enablethinking=False), seed42, eight responses per question, context 8192, and four learner plus four sampler GPUs. This uses DAPO asymmetric clipping, dynamic filtering/refill of mixed-correctness groups, global token-normalized policy loss, and a soft overlength reward. The exact released-code provenance and implemented equations are recorded below. Unlike the original paper's no-KL recipe, this controlled comparison uses the shared KL coefficient 0.01 against the frozen initial base. This is not an exact reproduction of the paper's scale or training recipe and makes no…

Fresh LoRA rank 16/alpha 32 on the same 2048-question split used by the STRIDE comparisons. Explicit nonthinking (enablethinking=False), seed42, eight responses per question, context 8192, and four learner plus four sampler GPUs. This uses DAPO asymmetric clipping, dynamic filtering/refill of mixed-correctness groups, global token-normalized policy loss, and a soft overlength reward. The exact released-code provenance and implemented equations are recorded below. Unlike the original paper's no-KL recipe, this controlled comparison uses the shared KL coefficient 0.01 against the frozen initial base. This is not an exact reproduction of the paper's scale or training recipe and makes no…

Model · Text generation

Decision-1-NF4

Worthify AI

A pre-quantized NF4 checkpoint in the Worthify Decision-1 family. Download 7.64 GB of weight shards and load the saved NF4 tensors directly. In three short usage checks on an RTX 3090, peak PyTorch-allocated inference memory was 7.26 GiB. We converted Decision-1 at cab55818cfb6f4163f2b2597bed31cceb6ba2030 with bitsandbytes NF4 and double quantization. This release adds no training. Decision-1 is a single-seed research preview derived from Gemma 4 12B; its original model card describes the training and BF16 evaluation. Give the model a text state, a question, and 2–16 described options. Read the next-token scores for the supplied option letters to select an option ID. Those scores are…

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations…