全站搜索
AI 模型 2119
- gemma4-coding-agent↗
S4MPL3BI4S
**Example usage**: - For text only LLMs: `llama-cli -hf S4MPL3BI4S/gemma4-coding-agent --jinja` - For multimodal models: `llama-mtmd-cli -hf S4MPL3BI4S/gemma4-coding-agent --jinja` To create an Ollama model from this vision model: 1. Place the `Modelfile` in the same directory as the finetuned bf16 merged model 3. Run: `ollama create model_name -f ./Modelfile` (Replace `model_name` with your desired name)
- Phi-3-medium-128k-instruct↗
Microsoft
🎉 **Phi-3.5**: [[mini-instruct]](https://huggingface.co/microsoft/Phi-3.5-mini-instruct); [[MoE-instruct]](https://huggingface.co/microsoft/Phi-3.5-MoE-instruct) ; [[vision-instruct]](https://huggingface.co/microsoft/Phi-3.5-vision-instruct) The Phi-3-Medium-128K-Instruct is a 14B parameters, lightweight, state-of-the-art open model trained with the Phi-3 datasets that includes both synthetic data and the filtered publicly available websites data with a focus on high-quality and reasoning dense properties. The model belongs to the Phi-3 family with the Medium version in two variants 4k and 128K which is the context length (in tokens) that it can support. The model has underwent a post-training process that incorporates both supervised fine-tuning and direct preference optimization for the instruction following and safety measures. When assessed against benchmarks testing common sense, language understanding, math, code, long context and logical reasoning, Phi-3-Medium-128K-Instruct showcased a robust and state-of-the-art performance among models of the same-size and next-size-up.
- eslav_PP-OCRv5_mobile_rec↗
PaddlePaddle
eslav_PP-OCRv5_mobile_rec is one of the PP-OCRv5_rec that are the latest generation text line recognition models developed by PaddleOCR team. It aims to efficiently and accurately support the recognition of East Slavic languages. The key accuracy metrics are as follow: **Note**: If any character (including punctuation) in a line was incorrect, the entire line was marked as wrong. This ensures higher accuracy in practical applications. You can also integrate the model inference of the text recognition module into your project. Before running the following code, please download the sample image to your local machine.
- samantha-hands-gguf↗
trommatic
- vietnamese-embedding↗
dangvantuan
**vietnamese-embedding** is the Embedding Model for Vietnamese language. This model is a specialized sentence-embedding trained specifically for the Vietnamese language, leveraging the robust capabilities of PhoBERT, a pre-trained language model based on the RoBERTa architecture. The model utilizes PhoBERT to encode Vietnamese sentences into a 768-dimensional vector space, facilitating a wide range of applications from semantic search to text clustering. The embeddings capture the nuanced meanings of Vietnamese sentences, reflecting both the lexical and contextual layers of the language. The model underwent a rigorous four-stage training and fine-tuning process, each tailored to enhance its ability to generate precise and contextually relevant sentence embeddings for the Vietnamese language. Below is an outline of these stages: - Dataset: ViNLI-SimCSE-supervised - Method: Trained using the SimCSE approach which employs a supervised contrastive learning framework. The model was optimized using Triplet Loss to effectively learn from high-quality annotated sentence pairs. - Dataset: XNLI-vn - Method: Continued fine-tuning using Multi-Negative Ranking Loss. This stage focused on improving the model's ability to discern and rank nuanced differences in sentence semantics. - Dataset: STSB-vn - Method: Fine-tuning specifically for the semantic textual similarity benchmark using Siamese BERT-Networks configured with the 'sentence-transformers' library. This stage honed the model's precision in capturing semantic similarity across various types of Vietnamese texts. - Dataset: STSB-vn with generate silver sample from gold sample - Method: Employed an advanced strategy using Augmented SBERT with Pair Sampling Strategies, integrating both Cross-Encoder and Bi-Encoder models. This stage further refined the embeddings by enriching the training data dynamically, enhancing the model's robustness and accuracy in understanding and processing complex Vietnamese language constructs. The performance is measured using Pearson and Spearman correlation: - On dev | Model | Pearson correlation | Spearman correlation | #params | | ------------- | ------------- | ------------- |------------
- Qwen-Image-2.1-Text-Encoder-GGUF↗
darrellbest
GGUF builds of the text encoder shipped in Qwen/Qwen-Image-2.1 (`text_encoder/`). That encoder is stock Qwen/Qwen3-VL-8B-Instruct: all 750 tensors match in name, shape and dtype, and every sampled tensor is byte-identical. So these files also work as a general Qwen3-VL-8B-Instruct GGUF. **Q6_K and Q4_K_M are not stock recipes.** Plain Q6_K visibly garbled rendered text, so both use: - an importance matrix computed on ~1500 real image prompts wrapped in the pipeline's own encoder template (`llama-imatrix --parse-special`); - the last 4 decoder layers (32-35) kept at Q8_0, because the pipeline reads the last layer's output directly; - token embeddings at Q8_0. **Recommendation:** Q8_0 if it fits. Q6_K is the smallest build that rendered text correctly in testing. Q4_K_M is usable, but at the same seed it changes composition and can misspell rendered text.
- granite-timeseries-ttm-r3↗
ibm-granite
Building on top of **TTM-R1** and **TTM-R2**, we introduce the next generation of TinyTimeMixer under the Granite time-series foundation model family — **Granite-TTM-R3**. This release incorporates several novel *tiny-neural architectural innovations* designed to push the limits of **accuracy in high-speed forecasting**, a critical requirement for real-world production deployments. **Granite-TTM-R3** is a family of pretrained models supporting multiple real-world forecasting scenarios: - **Zero-shot forecasting** across unseen datasets - **Few-shot adaptation** effective with as few as ~1K samples - **Full fine-tuning** for domain-specific optimization - **Multivariate time-series forecasting** - **Exogenous / control variable integration** - **High-throughput batch inference** for production systems
- imajev-9b-GGUF↗
mindchain
The same page, then the same page blackened. Same model, same question, same temperature. **The distribution moves from A 0.482 to D 0.887** — the image reaches the forward pass. This is the one test that separates a vision decision model from a text decision model that has been given a picture it cannot see: send an obviously different image and check whether the answer changes. Twenty seconds of work, and on this hardware it is what proved the Pascal vision tower runs (`--no-mmproj-offload`) and what caught every wrong turn on the way. The distributions on this card were measured with the **task in the state**. That matters more than it sounds. Asked "does this page fulfil its task?" without the task, the same model gave nearly flat answers — spread 0.089 across twenty pages instead of 0.176 — because it was guessing, and guessing looks plausible. On this hardware a decision model can only judge a page against a task it has actually been given.
- North-Mini-Code-1.0-GGUF↗
unsloth
Unsloth Dynamic 2.0 achieves superior accuracy & outperforms other leading quants. These are GGUF quants of North-Mini-Code-1.0. The model uses the `cohere2moe` architecture, which is not in a stock llama.cpp release yet. Until llama.cpp PR #24260 is merged, build llama.cpp from that PR branch to load these files. Once the PR lands in a release, these same GGUFs will run on stock llama.cpp with no re-download, because they already declare `general.architecture = cohere2moe`. The binaries are written to `build/bin/` (`llama-cli`, `llama-server`, `llama-quantize`).
- cartographer-ecommerce-reranker-MiniLM-L6-v2↗
2013khansohail
Cartographer is a small, fast cross-encoder that scores how well a shopping query matches a product. It takes a `(query, product)` pair and returns a single relevance score for reranking a candidate list. It is a drop-in upgrade for the general-purpose reranker it is built on: **same size, same speed, measurably better on product search.** It was fine-tuned from `cross-encoder/ms-marco-MiniLM-L-6-v2` on Amazon's public **Shopping Queries Dataset (ESCI)**. The base model was trained on web-search passages; product listings are not web text, so a domain-tuned model of the **same 22.7M-parameter footprint** does meaningfully better. Evaluated on the held-out ESCI test split (US / English): **8,956 queries, 181,701 (query, product) pairs.** Both models scored at `max_length=256`, identical settings : the comparison is apples-to-apples. The headline metric is nDCG (graded relevance), which is what the E/S/C/I labels are for.