Brief-details: A 3B parameter LLM specialized in function calling tasks, achieving near GPT-4 performance in API interaction and parameter identification with 57.69% overall accuracy on BFCL benchmark
Brief-details: Meta's Llama-3.1-405B-FP8 is a large language model with 405B parameters using 8-bit floating point quantization, part of Meta's advanced Llama 3 series.
BRIEF DETAILS: BlackSheep-Qwen-14B-i1-GGUF is a collection of GGUF quantized versions of the BlackSheep-Qwen-14B model, offering various compression levels from 3.7GB to 12.2GB with imatrix implementations.
BRIEF-DETAILS: Citrus1.0-Qwen-72B is a medical LLM built on Qwen-72B that emulates expert clinical reasoning pathways for enhanced medical decision support and diagnosis.
Brief-details: A quantized 3B parameter language model offering various GGUF formats optimized for different size/quality trade-offs, with IQ4_XS and Q4_K_M being recommended variants.
BRIEF DETAILS: A 3B parameter GGUF-quantized language model with multiple compression variants, offering flexible trade-offs between size (1.5GB-6.5GB) and quality.
Brief Details: Vintix - A 332M parameter multi-task action model utilizing in-context reinforcement learning, trained on robotics datasets with 20 layers and 8192 sequence length.
BRIEF-DETAILS: A merged 7B parameter LLM based on Qwen2.5, combining multiple models and LoRAs for enhanced performance. Achieves 36.22 avg score on OpenLLM benchmarks.
BRIEF DETAILS: 494M parameter GGUF quantized model, offering multiple compression options from Q2 to Q8, optimized for efficient deployment with sizes ranging 0.4-1.1GB
Brief-details: A quantized version of TEST2-Q2.5-Lenned-14B optimized for efficiency with multiple GGUF variants, offering various size/quality tradeoffs from 3.7GB to 12.2GB
BRIEF-DETAILS: Cable is a context-aware bias model for length extrapolation in transformers, achieving better performance than traditional models with minimal overhead
Brief Details: A specialized 2.1B parameter language model enhanced for creative writing through directional enhancement technique, preserving core abilities while amplifying creative capabilities.
BRIEF-DETAILS: A quantized version of Qwen2.5-Monte-7B offering multiple GGUF variants for efficient deployment, with sizes ranging from 3.1GB to 15.3GB
BRIEF-DETAILS: T-lite-it-1.0-GGUF is a quantized model offering various compression levels from 3.1GB to 15.3GB, with recommended Q4_K variants balancing speed and quality.
BRIEF DETAILS: Optimized GGUF quantization of Qwen2.5-3B model with multiple compression variants, ranging from 1.0GB to 2.9GB, featuring imatrix and static quantization options.
Brief Details: Llama-3.1-8B-Instruct-Uncensored-DeLMAT is a modified 8B parameter LLaMA model using custom DeLMAT training for reduced content filtering.
BRIEF-DETAILS: 12B parameter GGUF-quantized language model optimized for human-like responses, offering multiple quantization options from 4.9GB to 13.1GB with varying quality-size tradeoffs.
Brief-details: A quantized version of Crazy-Qwen2-7b offering multiple GGUF variants (2.7B-15.3GB) with different quality-size tradeoffs. Features both standard and IQ quantization options.
BRIEF-DETAILS: A 12B parameter GGUF-quantized Mistral model optimized for story writing, offering multiple quantization options from 4.9GB to 13.1GB with varying quality-size tradeoffs.
Brief-details: Gemma Coder 9B optimized GGUF model with various quantization options (2.5GB-7.7GB), offering flexible trade-offs between size, speed & quality for coding tasks.
Brief-details: OopsHusBot-3B is a 3B parameter LLaMA-based model fine-tuned for relationship communication assistance, featuring specialized modes for husband-wife dialogue optimization and conflict resolution.
