Brief Details: RoBERTa-based news classifier fine-tuned on AG News dataset. FP16-quantized for efficient deployment. 94.3% accuracy across 4 news categories.
BRIEF-DETAILS: 14B parameter GGUF quantized LLM with multiple compression options (Q2-Q8), optimized for efficient deployment with file sizes ranging from 5.9GB to 15.8GB
Brief Details: A GGUF quantized version of Tucana-Opus-14B featuring multiple compression variants, optimized for different size/performance tradeoffs at 14B parameters.
BRIEF-DETAILS: GGUF quantized version of Rune-14b offering multiple compression options (Q2-Q8). Features optimized versions for different size/performance trade-offs.
BRIEF DETAILS: Dark Llama GGUF model with multiple quantization options (2.1GB-6.7GB), optimized for various performance/size tradeoffs. Features IQ matrix quantization.
BRIEF-DETAILS: A 7B parameter GGUF-quantized model offering multiple compression variants from 3.1GB to 15.3GB, with recommended Q4_K variants balancing speed and quality.
BRIEF DETAILS: 7B parameter GGUF-quantized model optimized for code tasks, offering multiple quantization options from 3.1GB to 15.3GB with recommended Q4_K variants for balanced performance.
Brief Details: Wan2.1_i2v_720p_nf4 is an image-to-video conversion model created by e-n-v-y, optimized for 720p resolution with nf4 architecture modifications
Brief-details: TherapyZ-Llama-3-8B-GGUF is a quantized version of Llama 3 8B model with multiple compression options, optimized for therapeutic applications
Brief-details: A 7B parameter instruction-tuned merged model combining multiple Qwen2.5 variants, featuring strong performance in reasoning tasks with 36.14 avg benchmark score
BRIEF DETAILS: 8B parameter Llama-based model optimized for GGUF format with multiple quantization options (2-8 bit). Offers various performance/size tradeoffs.
BRIEF DETAILS: A GGUF quantized version of Impish QWEN 7B-1M offering multiple compression variants from 3.1GB to 15.3GB, with recommended Q4_K variants balancing speed and quality.
BRIEF-DETAILS: RNA foundation model trained on 42M non-coding RNA sequences with 650M parameters, offering sequence classification and embedding capabilities
Brief-details: TeluguSBERT model fine-tuned for sentence similarity tasks in Telugu language, based on BERT architecture and optimized for semantic textual analysis.
BRIEF: AraBERTv2 is a 543MB/136M parameter Arabic BERT model trained on 200M sentences (77GB), featuring improved preprocessing and vocabulary coverage for Arabic NLP tasks.
Brief-details: A comprehensive GGUF quantization of the Gemma 2B model with multiple compression options, optimized for different hardware setups and RAM constraints
Brief-details: Vision transformer-based NSFW image classifier achieving 93.16% accuracy. Fine-tuned on 28k images with 5 categories: drawings, hentai, neutral, porn, and sexy.
Brief-details: Powerful vision transformer model with 99.3M params, focused on spatial attention design. Trained on ImageNet-1k, optimized for image classification and feature extraction.
BRIEF-DETAILS: Czech-English semantic embedding model by Seznam using SimCSE fine-tuning on dist-mpnet-paracrawl, optimized for similarity and retrieval tasks.
BRIEF DETAILS: A multilingual embedding model (494M params) based on Qwen2-0.5B, achieving SOTA performance on MTEB (64.94) and C-MTEB (64.13) benchmarks.
BRIEF-DETAILS: Qwen2.5-72B-Instruct-4bit: MLX-optimized 72B parameter instruction-tuned model, 4-bit quantized for efficient deployment and inference
