Brief Details: A powerful text-to-image model with 142k+ downloads, specialized in photorealistic generation with strong performance in portraits and detailed scenes.
Brief-details: A speech emotion recognition model based on Wav2vec 2.0, fine-tuned on MSP-Podcast dataset. 165M parameters, predicts arousal, dominance, and valence from audio.
Brief Details: French sentence embedding model based on CamemBERT-Large (337M params), achieving 85.9% Pearson correlation on STS benchmark. Optimized for semantic similarity.
Brief Details: Optimized 11B parameter vision-language model using FP8 quantization, supporting 8 languages with 50% reduced memory footprint for efficient deployment
Brief-details: A lightweight emotion classification model based on DistilBERT, offering 93.8% accuracy for detecting 6 emotions while being 40% smaller than BERT
Brief Details: TrOCR small-printed: 61.4M parameter OCR model for printed text recognition. Uses transformer-based vision encoder-decoder architecture.
Brief-details: Compact BERT variant (4 layers, 256 hidden) optimized for efficient pre-training and NLI tasks. MIT licensed with 145K+ downloads.
Brief Details: A powerful 40B parameter LLM trained on 1,000B tokens, optimized for inference with FlashAttention and multiquery architecture under Apache 2.0 license.
Brief Details: Inception-ResNet-v2 model with 55.9M params, optimized for ImageNet classification. Features hybrid architecture combining Inception modules with residual connections.
Brief Details: Persian NER model based on BERT architecture, specialized in recognizing 7 entity types with 93.40% F1 score on PEYMA dataset. Apache 2.0 licensed.
Brief Details: DialoGPT-large - Microsoft's state-of-the-art conversational AI model trained on 147M Reddit dialogues, offering human-like response generation capabilities.
Brief Details: Ultravox v0.3 is an 8.06B parameter multimodal Speech LLM combining Llama3.1-8B-Instruct and Whisper-small for speech/text processing with MIT license.
BRIEF DETAILS: Ukrainian speech recognition model based on wav2vec2-xls-r-300m with 315M parameters, achieving 4.63% WER on Common Voice 10.0 test set.
Brief Details: SapBERT model for biomedical entity representations, trained on UMLS with PubMedBERT base. 109M params, optimized for medical entity linking.
Brief Details: PEGASUS-XSUM is a powerful abstractive summarization model by Google, achieving 46.86 ROUGE-1 scores on XSUM dataset with strong performance across multiple domains.
Brief Details: MobileNetV3-Large model pretrained on ImageNet-21k-P and fine-tuned on ImageNet-1k. Efficient architecture with 5.51M parameters optimized for mobile devices.
Brief Details: Qwen2-1.5B-Instruct is a 1.54B parameter instruction-tuned language model with impressive performance in reasoning, coding, and multilingual tasks.
Brief-details: Qwen1.5-7B is a powerful 7.72B parameter transformer-based language model with 32K context length support, offering improved multilingual capabilities and enhanced performance.
Brief-details: 76B parameter multimodal LLM combining InternViT vision and Llama3 language models, offering strong performance in image, video understanding and grounding tasks.
Brief-details: A powerful text-to-image model focused on ultra-realistic image generation, particularly excelling in portrait creation with over 151K downloads and optimized for photorealistic outputs.
Brief-details: Mistral-7B-Instruct-v0.3-AWQ is a 4-bit quantized version of Mistral-7B-Instruct, optimized for efficient inference while maintaining performance using AWQ technology.
