전체 검색

AI 모델 2119

• chilloutmix↗

  stablediffusionapi

  --- license: creativeml-openrail-m tags: - stablediffusionapi.com - stable-diffusion-api - text-to-image - ultra-realistic pinned: true --- Coding in PHP/Node/Java etc? Have a look at docs for more code examples: View docs payload = json.dumps({ "key": "", "model_id": "chilloutmix", "prompt": "actual 8K portrait photo of gareth person, portrait, happy colors, bright eyes, clear eyes, warm smile, smooth soft skin, big dreamy eyes, beautiful intricate colored hair, symmetrical, anime wide eyes, soft lighting, detailed face, by makoto shinkai, stanley artgerm lau, wlop, rossdraws, concept art, digital painting, looking into camera", "negative_prompt": "painting, extra fingers, mutated hands, poorly drawn hands, poorly drawn face, deformed, ugly, blurry, bad anatomy, bad proportions, extra limbs, cloned face, skinny, glitchy, double torso, extra arms, extra hands, mangled fingers, missing lips, ugly face, distorted face, extra legs, anime", "width": "512", "height": "512", "samples": "1", "num_inference_steps": "30", "safety_checker": "no", "enhance_prompt": "yes", "seed": None, "guidance_scale": 7.5, "multi_lingual": "no", "panorama": "no", "self_attention": "no", "upscale": "no", "embeddings": "embeddings_model_id", "lora": "lora_model_id", "webhook": None, "track_id": None })

• ecaresnet50t.a1_in1k↗

  timm

  This model features: * ReLU activations * tiered 3-layer stem of 3x3 convolutions with pooling * 2x2 average pool + 1x1 convolution shortcut downsample * Efficient Channel Attention Recipe details: * ResNet Strikes Back `A1` recipe * LAMB optimizer with BCE loss * Cosine LR schedule with warmup

• pythia-14m-deduped↗

  EleutherAI

  > **Note:** Pythia-14M-deduped was trained after the original Pythia suite at the request of interpretability researchers who wanted a smaller model with the same training setup. It uses the same data, tokenizer, hyperparameter conventions, and checkpoint schedule as the rest of the Pythia suite. Unlike the original Pythia models (70M--12B), Pythia-14M-deduped was only trained on the deduplicated Pile and does not have a separate non-deduplicated variant. The primary intended use of Pythia is research on the behavior, functionality, and limitations of large language models. This suite is intended to provide a controlled setting for performing scientific experiments. We provide 154 checkpoints per model: - Initial `step0` - 10 log-spaced checkpoints `step{1,2,4...512}` - 143 evenly-spaced checkpoints from `step1000` to `step143000` These checkpoints are hosted on Hugging Face as branches. Note that branch `143000` corresponds exactly to the model checkpoint on the `main` branch of each model.

• gemma-4-E4B-it-Assamese-Updesh-GGUF↗

  mradermacher

• sam3↗

  tiny-random

• Mistral-Medium-3.5-128B-GGUF↗

  unsloth

  We worked with Mistral to fix Mistral Medium 3.5 inference issues affecting some implementations (**not related to Unsloth** or our quants). The issue came from a YaRN parsing quirk in implementations like transformers and llama.cpp. Setting `mscale_all_dim` from `1` to `0` fixes it, including the model forgetting previous conversations. Mistral has now pushed these fixes to their official repo. Mistral Medium 3.5 is our first flagship merged model. It is a dense 128B model with a 256k context window, handling instruction-following, reasoning, and coding in a single set of weights. Mistral Medium 3.5 replaces its predecessor Mistral Medium 3.1 and Magistral in Le Chat. It also replaces Devstral 2 in our coding agent Vibe. Concretely, expect better performance for instruct, reasoning and coding tasks in a new unified model in comparison with our previous released models. Reasoning effort is configurable per request, so the same model can answer a quick chat reply or work through a complex agentic run. We trained the vision encoder from scratch to handle variable image sizes and aspect ratios.

• GLM-5.2-FP8↗

  zai-org

  We're introducing GLM-5.2, our latest flagship model for long-horizon tasks. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on a **solid 1M-token context**. GLM-5.2's new capabilities include: - **Solid 1M Context:** A solid 1M-token context that stably sustains long-horizon work - **Advanced Coding with Flexible Effort**: Stronger coding capabilities with multiple thinking effort levels to balance performance and latency - **Improved Architecture**: We propose IndexShare, which reuses the same indexer across every four sparse attention layers, reducing per-token FLOPs by 2.9× at a 1M context length. We also improve GLM-5.2’s MTP layer for speculative decoding, increasing the acceptance length by up to 20% - **Pure Open**: An MIT open-source license — no regional limits, technical access without borders GLM-5.2 supports deployment with the following frameworks. Feel free to try them out: - SGLang (v0.5.13.post1+) — see cookbook - vLLM (v0.23.0+) — see recipes - Transformers (v0.5.12+) — see transformers docs - KTransformers (v0.5.12+) — see tutorial - Unsloth (v0.1.47-beta+) — see guide - For deployment on the `Ascend NPU` platform, inference frameworks such as vLLM-Ascend, xLLM and SGLang are supported — see here.

• wav2vec2-base-vietnamese↗

  dragonSwing

  Fine-tuned dragonSwing/wav2vec2-base-pretrain-vietnamese on Vietnamese Speech Recognition task using 100h labelled data from VSLP dataset. When using this model, make sure that your speech input is sampled at 16kHz. The model can be used directly (without a language model) as follows:

• Qwen3.5-9B-GGUF-MoQ↗

  w-ahmad

  >MoQ (Mixture of Quants) is a smart way to shrink AI models without losing their "brainpower." Unlike old methods that treat every part of the model the same, MoQ identifies the most important parts and keeps them high-quality, while heavily compressing the rest to save space.****Stop settling for uniform bitrates.** Standard quantization is a relic of the past, treating vital cognitive weights the same as redundant noise. ** --- The result? A model that punches significantly above its weight class. Benjamin Marie evaluated MoQ GGUFs ("Mixture of Quants") against Unsloth Dynamic (UD) quants, focusing on low-bit versions below 4 bits on average — the range where GGUF models typically struggle most. Results: At similar bits-per-weight (Bpw), MoQ outperforms Unsloth Dynamic quants by ~10% on benchmarks, while also being roughly 2× more token-efficient on average. --- | Folder Link | BPW | Total Size | Description | | :--- | :---: | :---: | :--- | | 📂 **BF16** | **16** | **17.92 GB** | 📂 **F16** | **16** | **17.92 GB** | 📂 **MoQ-Quants-Latest** | **3.2** | **3.58 GB** | 📂 **MoQ-Quants-Latest** | **3.6** | **4.03 GB** | 📂 **MoQ-Quants-Latest** | **3.8** | **4.22 GB** | 📂 **MoQ-Quants-Latest** | **4.1** | **4.64 GB** | 📂 **MoQ-Quants-Latest** | **4.3** | **4.84 GB** | 📂 **MoQ-Quants-Latest** | **4.6** | **5.11 GB** | 📂 **MoQ-Quants-Latest** | **4.8** | **5.37 GB** | 📂 **MoQ-Quants-Latest** | **4.9** | **5.49 GB** | 📂 **MoQ-Quants-Latest** | **5.1** | **5.75 GB** | 📂 **MoQ-Quants-Latest** | **5.3** | **5.92 GB** | 📂 **MoQ-Quants-Latest** | **6.6** | **7.36 GB** | 📂 **MoQ-mmproj** | **16.0** | **0.92 GB**

• needle3↗

  Cactus-Compute

  A foundation model for mobiles, wearables, robots, smart home, automotive and microcontrollers. The whole model is a single 8-29 MB file, and we trade general chat capacity to beat models 10x its size on mobile tool calls and match 2-3x bigger models on extraction. - **Tool calls**: given the functions your app exposes, Needle picks the right ones and fills every argument from what the user said. Ask for two things and you get two calls in order; ask for something no tool covers and you get an empty list, not a guess. - **Structured extraction**: declare a shape, hand over messy text, get typed fields back: an invoice, a booking, a notification, a form. The decode grammar guarantees the output parses, and extraction generalises to classification. - **Text embedding**: the same model returns a vector for a sentence, so an app can search, match and route locally. The same engine also runs Whistle, our 16.9 MB speech-to-text model in seven languages. Load `whistle.cact` beside `needle3.cact` and the one binary transcribes a clip on its own, or answers it against your tools in a single call and returns the calls and the transcript in one JSON object.