# Multi-task evaluation

A robot strategy model trained based on LeRobot for the task of picking up the power gem.

GIST-Embedding-v0 is a sentence embedding model based on sentence-transformers, mainly used for sentence similarity calculation and feature extraction tasks.

INT4 weight-quantized version of DeepSeek-R1, reducing GPU memory and disk space requirements by approximately 50% while maintaining original model performance.

Latxa is a 70B-parameter Basque language large language model based on Llama-3.1, fine-tuned with instructions and FP8 quantization, specifically optimized for Basque

Latxa 3.1 70B Instruct is an instruction-tuned version based on Llama-3.1 (Instruct), specifically optimized for Basque language, demonstrating excellent performance on multiple Basque benchmarks.

GTE-small is an efficient sentence embedding model based on the thenlper/gte-small foundation model, focusing on sentence similarity tasks.

Lamarckvergence-14B is a pre-trained language model merged via mergekit, combining Lamarck-14B-v0.7 and Qwenvergence-14B-v12-Prose-DS. It ranks first among models with fewer than 15B parameters on the Open LLM Leaderboard.

LENS-4000 is a transformer-based text embedding model focused on feature extraction and sentence similarity calculation, excelling in multiple text classification tasks.

Stella 400M is a medium-scale English text processing model focused on classification and information retrieval tasks.

A Turkish large language model developed by WiroAI, fine-tuned based on Llama-3.1-8B-Instruct, specializing in Turkish text generation and dialogue tasks.

Bielik-11B-v2 is a generative text model with 11 billion parameters, specifically developed and trained for Polish language text. It is initialized based on Mistral-7B-v0.2 and trained on 400 billion tokens.

Meltemi 7B Instruct v1.5 is a Greek instruction fine-tuned large language model improved based on Mistral 7B, focusing on Greek natural language processing tasks.

A large language model developed in the third iteration using the Self-Play Preference Optimization method based on the Meta-Llama-3-8B-Instruct architecture.

Jina Embeddings V2 Base is a sentence embedding model optimized for Chinese, which can convert text into high-dimensional vector representations for calculating sentence similarity and feature extraction.

OpenELM is a set of open-source efficient language models that use a hierarchical scaling strategy to efficiently allocate parameters in each layer of the Transformer model, thereby improving model accuracy.

Dmeta-embedding-zh-small is a model that performs excellently in multiple natural language processing tasks and is particularly suitable for Chinese text processing.

vortex-3b is a 2.78 billion parameter causal language model developed by OEvortex, based on the Pythia-2.8b model and fine-tuned on the Vortex-50k dataset.

bge_micro is a sentence embedding model based on sentence-transformers, focusing on sentence similarity calculation and feature extraction tasks.

A large-scale Polish sentence transformation model based on the RoBERTa architecture, focusing on sentence similarity calculation and feature extraction tasks.

A Polish sentence embedding model based on RoBERTa architecture, focusing on sentence similarity calculation and feature extraction tasks.

mmlw-e5-small is a sentence transformer model for sentence similarity tasks, supporting Polish text processing.

This is a Polish sentence transformer model based on the KartonBERTa architecture, primarily used for sentence similarity calculation and feature extraction tasks.

stella-base-zh-v2 is a Chinese semantic similarity calculation model based on sentence transformers, supporting various text similarity tasks and evaluation benchmarks.

bge_micro is a lightweight model focused on sentence similarity calculation, suitable for various natural language processing tasks.

bge_micro is a lightweight sentence similarity calculation model based on transformer architecture, specifically designed for efficient feature extraction and sentence similarity tasks.

Puma-3B is a text generation model fine-tuned based on OpenLLaMA 3B V2. It is trained on the ShareGPT Hyperfiltered dataset and is suitable for various text generation tasks.

OpenLLaMA is an open-source reproduction of Meta AI's LLaMA large language model, offering pre-trained models with 3B, 7B, and 13B parameters

Polyglot-Ko is one of the Korean autoregressive language model series developed by EleutherAI's multilingual team, containing 1.3 billion parameters and specifically optimized for Korean.

AraBERT is an Arabic pre-trained language model based on the BERT architecture, optimized for Arabic natural language understanding tasks.

SGPT-5.8B is a sentence transformer model based on the weighted mean method, focusing on sentence similarity tasks, trained on the msmarco dataset and optimized with specb-bitfit technology

SGPT-2.7B is a sentence transformer model based on the weighted mean method, focusing on sentence similarity tasks, trained on the MSMARCO dataset with BitFit technology applied.

This is a Hugging Face transformers-compatible version conversion of the 125M parameter dense model from Artetxe et al.'s paper