diff --git a/assets/adobe.yaml b/assets/adobe.yaml index 1a8331af..ec346907 100644 --- a/assets/adobe.yaml +++ b/assets/adobe.yaml @@ -97,3 +97,25 @@ monthly_active_users: unknown user_distribution: unknown failures: unknown + +- type: dataset + name: CulturaX + organization: University of Oregon, Adobe + description: CulturaX is a substantial multilingual dataset with 6.3 trillion tokens in 167 languages, tailored for LLM development. + created_date: 2023-09-17 + url: https://arxiv.org/pdf/2309.09400 + datasheet: https://huggingface.co/datasets/uonlp/CulturaX + modality: text + size: 6.3 trillion tokens + sample: [] + analysis: none + dependencies: [mC4, OSCAR] + included: unknown + excluded: unknown + quality_control: unknown + access: open + license: mC4, OSCAR + intended_uses: '' + prohibited_uses: The data must not be utilized for malicious or harmful purposes towards humanity. + monitoring: unknown + feedback: https://huggingface.co/datasets/uonlp/CulturaX/discussions diff --git a/assets/ai2.yaml b/assets/ai2.yaml index c4d18597..4188c061 100644 --- a/assets/ai2.yaml +++ b/assets/ai2.yaml @@ -255,3 +255,25 @@ prohibited_uses: '' monitoring: unknown feedback: https://huggingface.co/allenai/OLMo-7B/discussions + +- type: dataset + name: MADLAD-400 + organization: AI2 + description: MADLAD-400 is a document-level multilingual dataset based on Common Crawl, covering 419 languages in total. + created_date: 2023-09-09 + url: https://arxiv.org/abs/2309.04662 + datasheet: https://huggingface.co/datasets/allenai/MADLAD-400 + modality: text + size: 3 trillion tokens + sample: [] + analysis: none + dependencies: [Common Crawl] + included: '' + excluded: '' + quality_control: '' + access: open + license: CC BY 4.0 + intended_uses: '' + prohibited_uses: '' + monitoring: unknown + feedback: https://huggingface.co/datasets/allenai/MADLAD-400/discussions diff --git a/assets/alibaba.yaml b/assets/alibaba.yaml index 2e900a97..1541b894 100644 --- a/assets/alibaba.yaml +++ b/assets/alibaba.yaml @@ -87,7 +87,7 @@ - type: model name: Qwen 1.5 - organization: Qwen Team + organization: Alibaba description: Qwen 1.5 is the next iteration in their Qwen series, consisting of Transformer-based large language models pretrained on a large volume of data, including web texts, books, codes, etc. @@ -141,3 +141,27 @@ prohibited_uses: '' monitoring: unknown feedback: https://huggingface.co/Qwen/Qwen1.5-MoE-A2.7B/discussions + +- type: model + name: SeaLLM v2.5 + organization: DAMO Academy, Alibaba + description: SeaLLM v2.5 is a multilingual large language model for Southeast Asian (SEA) languages. + created_date: 2024-04-12 + url: https://github.com/DAMO-NLP-SG/SeaLLMs + model_card: https://huggingface.co/SeaLLMs/SeaLLM-7B-v2.5 + modality: text; text + analysis: The model was evaluated on 3 benchmarks (MMLU for English, M3Exam (M3e) for English, Chinese, Vietnamese, Indonesian, and Thai, and VMLU for Vietnamese) and it outperformed GPT-3 and Vistral-7B-chat models across these benchmarks in the given languages. + size: 7B parameters + dependencies: [Gemma] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: Despite efforts in red teaming and safety fine-tuning and enforcement, the creators suggest, developers and stakeholders should perform their own red teaming and provide related security measures before deployment, and they must abide by and comply with local governance and regulations. + access: open + license: + explanation: License can be found at https://huggingface.co/SeaLLMs/SeaLLM-13B-Chat/blob/main/LICENSE + value: custom + intended_uses: The model is intended for multilingual tasks such as knowledge retrieval, math reasoning, and instruction following. Also, it could be used to provide multilingual assistance. + prohibited_uses: The model should not be used in a way that could lead to inaccurate, misleading or potentially harmful generation. Users should comply with local laws and regulations when deploying the model. + monitoring: unknown + feedback: https://huggingface.co/SeaLLMs/SeaLLM-7B-v2.5/discussions diff --git a/assets/apple.yaml b/assets/apple.yaml index ab8b26a0..4462a34d 100644 --- a/assets/apple.yaml +++ b/assets/apple.yaml @@ -22,3 +22,24 @@ prohibited_uses: '' monitoring: '' feedback: none +- type: model + name: OpenELM + organization: Apple + description: OpenELM is a family of Open-source Efficient Language Models. It uses a layer-wise scaling strategy to efficiently allocate parameters within each layer of the transformer model, leading to enhanced accuracy. + created_date: 2024-04-24 + url: https://machinelearning.apple.com/research/openelm + model_card: https://huggingface.co/apple/OpenELM-3B-Instruct + modality: text; text + analysis: The models were evaluated in terms of zero-shot, LLM360, and OpenLLM leaderboard results. + size: 3B parameters + dependencies: [RefinedWeb, The Pile, RedPajama-Data, Dolma, CoreNet library] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: unknown + access: open + license: Apple + intended_uses: To empower and enrich the open research community by providing access to state-of-the-art language models. + prohibited_uses: No explicit prohibited uses stated, though it is noted that users should undertake thorough safety testing. + monitoring: none + feedback: https://huggingface.co/apple/OpenELM-3B-Instruct/discussions diff --git a/assets/cohere.yaml b/assets/cohere.yaml index bae1e048..7b9074df 100644 --- a/assets/cohere.yaml +++ b/assets/cohere.yaml @@ -543,3 +543,24 @@ prohibited_uses: '' monitoring: unknown feedback: https://huggingface.co/datasets/CohereForAI/aya_dataset/discussions +- type: model + name: Rerank 3 + organization: Cohere + description: Rerank 3 is a new foundation model for efficient enterprise search and retrieval with 4k context length. + created_date: 2024-04-11 + url: https://cohere.com/blog/rerank-3 + model_card: none + modality: text; text + analysis: Evaluated on code retrieval and data retrieval capabilities, with improvements compared to the standard in both. + size: unknown + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: limited + license: unknown + intended_uses: Efficient enterprise search and retrieval. + prohibited_uses: '' + monitoring: unknown + feedback: none diff --git a/assets/eleutherai.yaml b/assets/eleutherai.yaml index 320e67a0..2af747be 100644 --- a/assets/eleutherai.yaml +++ b/assets/eleutherai.yaml @@ -296,3 +296,24 @@ prohibited_uses: '' monitoring: none feedback: https://huggingface.co/datasets/EleutherAI/proof-pile-2/discussions +- type: model + name: Pile-T5 + organization: EleutherAI + description: Pile-T5 is a version of the broadly used T5 model, but improved to eliminate weaknesses such as the omission of crucial code-related tokens. It utilizes LLaMA tokenizer and is trained on the Pile, offering enhancements for finetuning on downstream tasks, particularly those involving code. + created_date: 2024-04-15 + url: https://blog.eleuther.ai/pile-t5/ + model_card: none + modality: text; text + analysis: The models were evaluated on SuperGLUE, CodeXGLUE, as well as MMLU and Bigbench Hard. Comparisons were made with T5v1.1 and found that Pile-T5 models performed better in most conditions. + size: unknown + dependencies: [The Pile, T5x, LLaMA, umT5] + training_emissions: unknown + training_time: 2 million steps + training_hardware: unknown + quality_control: '' + access: open + license: unknown + intended_uses: The model is aimed at downstream tasks that benefit from the encoder-decoder architecture. Particularly useful for tasks involving code. + prohibited_uses: unknown + monitoring: unknown + feedback: unknown diff --git a/assets/fuse.yaml b/assets/fuse.yaml new file mode 100644 index 00000000..2607a5e2 --- /dev/null +++ b/assets/fuse.yaml @@ -0,0 +1,22 @@ +--- +- type: model + name: FuseChat + organization: FuseAI + description: FuseChat is a powerful chat Language Learning Model (LLM) that integrates multiple structure and scale-varied chat LLMs using a fuse-then-merge strategy. The fusion is done using two stages + created_date: 2024-02-26 + url: https://arxiv.org/abs/2402.16107 + model_card: https://huggingface.co/FuseAI/FuseChat-7B-VaRM + modality: text; text + analysis: The FuseChat model was evaluated on MT-Bench which comprises 80 multi-turn dialogues spanning writing, roleplay, reasoning, math, coding, stem, and humanities domains. It yields an average performance of 66.52 with specific scores for individual domains available in the leaderboard results. + size: 7B parameters + dependencies: [Nous Hermes 2, OpenChat 3.5] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: none + access: open + license: Apache 2.0 + intended_uses: FuseChat is intended to be used as a powerful chat bot that takes in text inputs and provides text-based responses. It can be utilized in a variety of domains including writing, roleplay, reasoning, math, coding, stem, and humanities. + prohibited_uses: unknown + monitoring: unknown + feedback: https://huggingface.co/FuseAI/FuseChat-7B-VaRM/discussions diff --git a/assets/google.yaml b/assets/google.yaml index 64eef91f..c6dee979 100644 --- a/assets/google.yaml +++ b/assets/google.yaml @@ -1782,3 +1782,25 @@ here https://ai.google.dev/gemma/prohibited_use_policy monitoring: '' feedback: https://huggingface.co/google/gemma-7b/discussions +- type: model + name: Med-Gemini + organization: Google + description: Med-Gemini is a family of highly capable multimodal models that are specialized in medicine with the ability to seamlessly integrate the use of web search, and that can be efficiently tailored to novel modalities using custom encoders. + created_date: 2024-04-29 + url: https://arxiv.org/pdf/2404.18416 + model_card: none + modality: image, text; text + analysis: Evaluated Med-Gemini on 14 medical benchmarks spanning text, multimodal and long-context applications, establishing new state-of-the-art (SoTA) performance on 10 of them, and surpassing the GPT-4 model family on every benchmark where a direct comparison is viable. + size: unknown + dependencies: [Gemini, MultiMedBench] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: closed + license: unknown + intended_uses: To be used in areas of medical research including medical summarization, referral letter generation, and medical simplification tasks. + prohibited_uses: Unfit for real-world deployment in the safety-critical medical domain. + monitoring: '' + feedback: none + diff --git a/assets/huggingface.yaml b/assets/huggingface.yaml index 606f4ea1..75df5184 100644 --- a/assets/huggingface.yaml +++ b/assets/huggingface.yaml @@ -171,3 +171,47 @@ prohibited_uses: unknown monitoring: '' feedback: https://huggingface.co/datasets/HuggingFaceTB/cosmopedia/discussions +- type: model + name: Idefics2 + organization: Hugging Face + description: Idefics2 is a general multimodal model that takes as input arbitrary sequences of text and images, generating text responses. It has the capability to describe visual content, answer questions about images, perform basic arithmetic operations, create stories grounded in multiple images, and extract information from documents. + created_date: 2024-04-15 + url: https://huggingface.co/blog/idefics2 + model_card: https://huggingface.co/HuggingFaceM4/idefics2-8b + modality: image, text; text + analysis: The performance of Idefics2 has been evaluated on numerous benchmarks. It is top of its class size and competes with much larger models such as LLava-Next-34B and MM1-30B-chat. + size: 8B parameters + dependencies: [The Cauldron] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: The quality of the model has been ensured by training it on a mixture of openly available datasets and enhancing its OCR capabilities. Further improvements include manipulating images in their native resolutions and aspect ratios, better pre-trained backbones, and allowing for sub-image splitting. + access: open + license: Apache 2.0 + intended_uses: The model can be used for answering questions about images, describing visual content, creating stories grounded in multiple images, extracting information from documents, and performing basic arithmetic operations. + prohibited_uses: unknown + monitoring: unknown + feedback: https://huggingface.co/HuggingFaceM4/idefics2-8b/discussions +- type: dataset + name: The Cauldron + organization: Hugging Face + description: The Cauldron is an open compilation of 50 manually-curated datasets formatted for multi-turn conversations. + created_date: 2024-04-15 + url: https://huggingface.co/blog/idefics2 + datasheet: https://huggingface.co/datasets/HuggingFaceM4/the_cauldron + modality: image, text + size: 50 vision-language datasets + sample: [] + analysis: none + dependencies: + explanation: These are the datasets with the most tokens included; the full list of all 50 datasets can be found at https://huggingface.co/datasets/HuggingFaceM4/the_cauldron + value: [LNarratives, Rendered Text, WebSight, DaTikz] + included: '' + excluded: '' + quality_control: unknown + access: open + license: CC BY 4.0 + intended_uses: '' + prohibited_uses: '' + monitoring: unknown + feedback: https://huggingface.co/datasets/HuggingFaceM4/the_cauldron/discussions diff --git a/assets/konan.yaml b/assets/konan.yaml new file mode 100644 index 00000000..55d90d11 --- /dev/null +++ b/assets/konan.yaml @@ -0,0 +1,22 @@ +--- +- type: model + name: Konan LLM + organization: Konan + description: Konan LLM is a Large Language Model developed in-house by Konan Technology. Optimized for super-large AI training, it leverages high-quality, large-scale data and over 20 years of expertise in natural language processing. + created_date: 2023-09-17 + url: https://en.konantech.com/en/llm/konanllm + model_card: none + modality: text; text + analysis: none + size: 13B parameters + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: limited + license: unknown + intended_uses: Document generation, document review, Q&A, customer response scenarios. + prohibited_uses: '' + monitoring: '' + feedback: none \ No newline at end of file diff --git a/assets/ktai.yaml b/assets/ktai.yaml new file mode 100644 index 00000000..56251384 --- /dev/null +++ b/assets/ktai.yaml @@ -0,0 +1,22 @@ +--- +- type: model + name: Midm + organization: KT Corporation + description: Midm is a pre-trained Korean-English language model developed by KT. It takes text as input and creates text. The model is based on Transformer architecture for an auto-regressive language model. + created_date: 2023-10-31 + url: https://huggingface.co/KT-AI/midm-bitext-S-7B-inst-v1 + model_card: https://huggingface.co/KT-AI/midm-bitext-S-7B-inst-v1 + modality: text; text + analysis: unknown + size: 7B parameters + dependencies: [AI-HUB dataset, National Institute of Korean Language dataset] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: KT tried to remove unethical expressions such as profanity, slang, prejudice, and discrimination from training data. + access: open + license: CC-BY-NC 4.0 + intended_uses: It is expected to be used for various research purposes. + prohibited_uses: It cannot be used for commercial purposes. + monitoring: unknown + feedback: https://huggingface.co/KT-AI/midm-bitext-S-7B-inst-v1/discussions diff --git a/assets/lg.yaml b/assets/lg.yaml new file mode 100644 index 00000000..3ec7981c --- /dev/null +++ b/assets/lg.yaml @@ -0,0 +1,22 @@ +--- +- type: model + name: EXAONE 2.0 + organization: LG AI Research + description: EXAONE 2.0 is a multimodal artificial intelligence that can be used to help develop new materials and medicines. + created_date: 2023-07-19 + url: https://www.lgresearch.ai/exaone + model_card: none + modality: image, text; image, text + analysis: none + size: unknown + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: closed + license: unknown + intended_uses: '' + prohibited_uses: '' + monitoring: '' + feedback: none diff --git a/assets/meta.yaml b/assets/meta.yaml index 3fbe3520..a2908725 100644 --- a/assets/meta.yaml +++ b/assets/meta.yaml @@ -319,9 +319,9 @@ monitoring: '' feedback: '' - type: model - name: LLaMA 2 + name: Llama 2 organization: Meta - description: LLaMA 2 is an updated version of LLaMA trained on a new mix of publicly + description: Llama 2 is an updated version of LLaMA trained on a new mix of publicly available data. created_date: 2023-07-18 url: https://ai.meta.com/resources/models-and-libraries/llama/ @@ -338,13 +338,13 @@ license: explanation: The license can be found at https://ai.meta.com/resources/models-and-libraries/llama-downloads/ value: custom - intended_uses: LLaMA 2 is intended for commercial and research use in English. + intended_uses: Llama 2 is intended for commercial and research use in English. Tuned models are intended for assistant-like chat, whereas pretrained models can be adapted for a variety of natural language generation tasks. prohibited_uses: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. Use in any other way that is prohibited by the Acceptable Use Policy and Licensing - Agreement for LLaMA 2. + Agreement for Llama 2. monitoring: '' feedback: '' - type: model @@ -706,13 +706,13 @@ modality: text; code, text analysis: Evaluated on several code benchmarks like HumanEval and MBPP. size: 34B parameters (dense) - dependencies: [LLaMA 2] + dependencies: [Llama 2] training_emissions: 65.3 tCO2eq training_time: 400K GPU hours training_hardware: A100-80GB GPUs quality_control: '' access: open - license: LLaMA 2 + license: Llama 2 intended_uses: Code Llama and its variants is intended for commercial and research use in English and relevant programming languages. prohibited_uses: Use in any manner that violates applicable laws or regulations @@ -793,3 +793,27 @@ prohibited_uses: '' monitoring: none feedback: none + +- type: model + name: Llama 3 + organization: Meta + description: Llama 3 is the third generation of Meta AI's open-source large language model. It comes with pretrained and instruction-fine-tuned language models with 8B and 70B parameters that can support a broad range of use cases. + created_date: 2024-04-18 + url: https://llama.meta.com/llama3/ + model_card: https://github.com/meta-llama/llama3/blob/main/MODEL_CARD.md + modality: text; text + analysis: The models were evaluated based on their performance on standard benchmarks and real-world scenarios. These evaluations were performed using a high-quality human evaluation set containing 1,800 prompts covering multiple use cases. The models also went through red-teaming for safety, where human experts and automated methods were used to generate adversarial prompts to test for problematic responses. + size: 70B parameters + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: 2 custom-built Meta 24K GPU clusters + quality_control: Extensive internal and external testing for safety, and design of new trust and safety tools. + access: open + license: + explanation: Can be found at https://github.com/meta-llama/llama3/blob/main/LICENSE + value: Llama 3 + intended_uses: Llama 3 is intended for a broad range of use cases, including AI assistance, content creation, learning, and analysis. + prohibited_uses: unknown + monitoring: Extensive internal and external performance evaluation and red-teaming approach for safety testing. + feedback: Feedback is encouraged from users to improve the model, but the feedback mechanism is not explicitly described. diff --git a/assets/microsoft.yaml b/assets/microsoft.yaml index 0727b96d..646bc637 100644 --- a/assets/microsoft.yaml +++ b/assets/microsoft.yaml @@ -912,3 +912,24 @@ prohibited_uses: Any purposes other than research. monitoring: unknown feedback: https://huggingface.co/microsoft/Orca-2-13b/discussions +- type: model + name: Phi-3 Mini + organization: Microsoft + description: Phi-3 Mini is a 3.8 billion-parameter, lightweight, state-of-the-art open model trained using the Phi-3 datasets. + created_date: 2024-04-23 + url: https://azure.microsoft.com/en-us/blog/introducing-phi-3-redefining-whats-possible-with-slms/ + model_card: https://huggingface.co/microsoft/Phi-3-mini-128k-instruct + modality: text; text + analysis: The model has been evaluated against benchmarks that test common sense, language understanding, mathematics, coding, long-term context, and logical reasoning. The Phi-3 Mini-128K-Instruct demonstrated robust and state-of-the-art performance among models with fewer than 13 billion parameters. + size: 3.8B parameters + dependencies: [] + training_emissions: unknown + training_time: 7 days + training_hardware: 512 H100-80G GPUs + quality_control: The model underwent post-training processes viz. supervised fine-tuning and direct preference optimization to increase its capability in following instructions and aligning to safety measures. + access: open + license: MIT + intended_uses: The model's primary use cases are for commercial and research purposes that require capable reasoning in memory or compute constrained environments and latency-bound scenarios. It can also serve as a building block for generative AI-powered features. + prohibited_uses: The model should not be used for high-risk scenarios without adequate evaluation and mitigation techniques for accuracy, safety, and fairness. + monitoring: Issues like allocation, high-risk scenarios, misinformation, generation of harmful content and misuse should be monitored and addressed. + feedback: https://huggingface.co/microsoft/Phi-3-mini-128k-instruct/discussions diff --git a/assets/naver.yaml b/assets/naver.yaml index 79a21cc5..37cb740f 100644 --- a/assets/naver.yaml +++ b/assets/naver.yaml @@ -1,16 +1,16 @@ --- - type: model name: HyperCLOVA - organization: Naver + organization: NAVER description: HyperClova is an autoregressive language model created_date: explanation: The date the model paper was submitted to arxiv value: 2021-05-21 url: https://arxiv.org/abs/2109.04650 - model_card: '' + model_card: none modality: text; text analysis: '' - size: 82B parameters (dense) + size: 82B parameters dependencies: [] training_emissions: '' training_time: 130.4 days @@ -24,3 +24,25 @@ prohibited_uses: '' monitoring: '' feedback: '' + +- type: model + name: HyperCLOVA X + organization: NAVER + description: HyperCLOVA X is a family of large language models (LLMs) tailored to the Korean language and culture, along with competitive capabilities in English, math, and coding. + created_date: 2024-04-13 + url: https://arxiv.org/pdf/2404.01954 + model_card: none + modality: text; text + analysis: Evaluated on English and Korean benchmarks in comparison to open source English and multilingual LLMs, with HyperCLOVA X (closed) surpassing the models compared. + size: unknown + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: limited + license: unknown + intended_uses: '' + prohibited_uses: '' + monitoring: '' + feedback: none diff --git a/assets/ncsoft.yaml b/assets/ncsoft.yaml new file mode 100644 index 00000000..2bc86592 --- /dev/null +++ b/assets/ncsoft.yaml @@ -0,0 +1,24 @@ +--- +- type: model + name: VARCO-LLM + organization: NCSOFT + description: VARCO-LLM is NCSOFT’s large language model and is trained on English and Korean. + created_date: 2023-08-16 + url: https://github.com/ncsoft/ncresearch + model_card: none + modality: text; text + analysis: Boasts the highest performance among the Korean LLMs of similar sizes that have been released to date, according to internal evaluations. + size: 13B parameters + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: closed + license: + explanation: Can be found at https://github.com/ncsoft/ncresearch/blob/main/LICENSE.txt + value: custom + intended_uses: Developing various NLP-based AI services such as Q&A, chatbot, summarization, information extraction + prohibited_uses: '' + monitoring: '' + feedback: none diff --git a/assets/openbmb.yaml b/assets/openbmb.yaml index 180132ae..c961d875 100644 --- a/assets/openbmb.yaml +++ b/assets/openbmb.yaml @@ -71,3 +71,24 @@ prohibited_uses: '' monitoring: unknown feedback: https://huggingface.co/openbmb/MiniCPM-V/discussions +- type: model + name: Eurus + organization: OpenBMB + description: Eurus is a suite of large language models (LLMs) optimized for reasoning. + created_date: 2024-04-02 + url: https://arxiv.org/abs/2404.02078 + model_card: https://huggingface.co/openbmb/Eurus-70b-nca + modality: text; text + analysis: The model was comprehensively benchmarked across 12 tests covering five tasks. Eurus achieved the best overall performance among open-source models of similar sizes and even outperformed specialized models in many cases. + size: 70B parameters + dependencies: [Eurus SFT, UltraInteract, UltraFeedback] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: none + access: open + license: Apache 2.0 + intended_uses: The model can be used for reasoning tasks and is especially tailored for coding and math following specific prompts. + prohibited_uses: none + monitoring: unknown + feedback: https://huggingface.co/openbmb/Eurus-70b-nca/discussions diff --git a/assets/reka.yaml b/assets/reka.yaml index c338c46d..7dd3f20c 100644 --- a/assets/reka.yaml +++ b/assets/reka.yaml @@ -21,3 +21,24 @@ prohibited_uses: '' monitoring: unknown feedback: none +- type: model + name: Reka Core + organization: Reka + description: Reka Core is a frontier-class multimodal language model comparable to industry leaders. It has powerful capabilities including multimodal understanding (including images, videos, and audio), superb reasoning abilities, code generation, and multilinguality with proficiency in 32 languages. + created_date: 2024-04-15 + url: https://www.reka.ai/news/reka-core-our-frontier-class-multimodal-language-model + model_card: none + modality: audio, image, text, video; text + analysis: Reka Core was evaluated against leading models such as OpenAIs GPT-4, Claude-3 Opus, and Gemini Ultra on a variety of tasks and metrics including multimodal and human evaluation conducted by a third party. It was found to be competitive or even surpassing these models. + size: unknown + dependencies: [] + training_emissions: unknown + training_time: few months + training_hardware: thousands of GPUs + quality_control: '' + access: limited + license: unknown + intended_uses: Reka Core can be used in e-commerce, social media, digital content and video games, healthcare, robotics, and other industries for tasks that require multimodal understanding, coding, complex reasoning, and more. + prohibited_uses: unknown + monitoring: unknown + feedback: unknown diff --git a/assets/shanghai.yaml b/assets/shanghai.yaml index 5cb3d267..c549fbfe 100644 --- a/assets/shanghai.yaml +++ b/assets/shanghai.yaml @@ -127,3 +127,46 @@ prohibited_uses: '' monitoring: unknown feedback: none +- type: model + name: CosmicMan + organization: Shanghai AI Laboratory + description: CosmicMan is a text-to-image foundation model specialized for generating high-fidelity human images with meticulous appearance, reasonable structure, and precise text-image alignment. + created_date: 2024-04-01 + url: https://cosmicman-cvpr2024.github.io/ + model_card: none + modality: text; image + analysis: The model was compared with SOTAs and has shown good performance in generating high-quality human images. + size: unknown + dependencies: [CosmicMan-HQ 1.0] + training_emissions: unknown + training_time: 1 week + training_hardware: 32 80G NVIDIA A100 GPUs + quality_control: The quality control measures taken include modeling the relationship between dense text descriptions and image pixels in a decomposed manner and enforcing attention refocusing without adding extra modules. + access: open + license: unknown + intended_uses: The model is intended to generate high-quality, photorealistic human images from text descriptions. Applications include avatar generation and potentially virtual reality and video game character creation. + prohibited_uses: unknown + monitoring: unknown + feedback: unknown + +- type: dataset + name: CosmicMan-HQ 1.0 + organization: Shanghai AI Laboratory + description: CosmicMan-HQ 1.0 is a large-scale dataset with 6 million high-quality, real-world human images. + created_date: 2024-04-28 + url: https://arxiv.org/pdf/2404.01294 + datasheet: none + modality: image + size: 6 million images + sample: [] + analysis: Compared to other human image datasets on data quantity, image quality, and annotations. + dependencies: [] + included: '' + excluded: '' + quality_control: unknown + access: open + license: unknown + intended_uses: '' + prohibited_uses: '' + monitoring: unknown + feedback: none diff --git a/assets/skt.yaml b/assets/skt.yaml new file mode 100644 index 00000000..b2c2d46a --- /dev/null +++ b/assets/skt.yaml @@ -0,0 +1,22 @@ +--- +- type: model + name: A.X + organization: SK Telecom + description: A.X is SK Telecom's proprietary LLM, which has been trained on the Korean language. + created_date: 2023-09-26 + url: https://www.sktelecom.com/en/press/press_detail.do?idx=1582 + model_card: none + modality: text; text + analysis: none + size: 39B parameters + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: closed + license: unknown + intended_uses: '' + prohibited_uses: '' + monitoring: '' + feedback: none \ No newline at end of file diff --git a/assets/stability.yaml b/assets/stability.yaml index 8695e197..128631bf 100644 --- a/assets/stability.yaml +++ b/assets/stability.yaml @@ -263,3 +263,25 @@ Policy. monitoring: unknown feedback: https://huggingface.co/stabilityai/sv3d/discussions + +- type: model + name: Stable Audio 2.0 + organization: Stability AI + description: Stable Audio 2.0 sets a new standard in AI-generated audio, producing high-quality, full tracks with coherent musical structure up to three minutes in length at 44.1kHz stereo. + created_date: 2024-04-03 + url: https://stability-ai.squarespace.com/news/stable-audio-2-0 + model_card: none + modality: audio, text; audio + analysis: none + size: unknown + dependencies: [AudioSparx] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: To protect creator copyrights, for audio uploads, Stability AI partners with Audible Magic to use their content recognition (ACR) technology to power real-time content matching and prevent copyright infringement. Opt-out requests were honored during the training phase. + access: open + license: unknown + intended_uses: It can be used to generate melodies, backing tracks, stems, and sound effects. + prohibited_uses: Uploading copyrighted material for transformation. + monitoring: Advanced content recognition is used to maintain compliance and prevent copyright infringement. + feedback: none diff --git a/assets/tokyo.yaml b/assets/tokyo.yaml new file mode 100644 index 00000000..7930e47e --- /dev/null +++ b/assets/tokyo.yaml @@ -0,0 +1,24 @@ +--- +- type: model + name: Aurora-M + organization: Tokyo Institute of Technology, MIT-IBM Watson Lab, Sapienza University of Rome + description: Aurora-M is a 15B parameter multilingual open-source model trained on English, Finnish, Hindi, Japanese, Vietnamese, and code. + created_date: 2024-04-23 + url: https://arxiv.org/pdf/2404.00399 + model_card: none + modality: text; text + analysis: Evaluated on all language datasets compared to similarly sized SOTA models, with Aurora-M achieving strong performance in most. + size: 15B parameters + dependencies: [StarCoderPlus] + training_emissions: + explanation: The training process operated entirely on 100% hydro-powered energy and included waste heat recycling. + value: unknown + training_time: 48 days + training_hardware: LUMI supercomputer, using 128 AMD MI250X GPUs + quality_control: '' + access: open + license: unknown + intended_uses: '' + prohibited_uses: '' + monitoring: unknown + feedback: none \ No newline at end of file diff --git a/assets/xai.yaml b/assets/xai.yaml index 8e0dc631..14dc7b20 100644 --- a/assets/xai.yaml +++ b/assets/xai.yaml @@ -23,3 +23,24 @@ prohibited_uses: none monitoring: unknown feedback: none +- type: model + name: Grok-1.5V + organization: xAI + description: Grok-1.5V is a first-generation multimodal model which can process a wide variety of visual information, including documents, diagrams, charts, screenshots, and photographs. + created_date: 2024-04-12 + url: https://x.ai/blog/grok-1.5v + model_card: none + modality: image, text; text + analysis: The model is evaluated in a zero-shot setting without chain-of-thought prompting. The evaluation domains include multi-disciplinary reasoning, understanding documents, science diagrams, charts, screenshots, photographs and real-world spatial understanding. The model shows competitive performance with existing frontier multimodal models. + size: unknown + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: limited + license: unknown + intended_uses: Grok-1.5V can be used for understanding documents, science diagrams, charts, screenshots, photographs. It can also translate diagrams into Python code. + prohibited_uses: unknown + monitoring: unknown + feedback: none diff --git a/js/main.js b/js/main.js index f0faca25..03c72139 100644 --- a/js/main.js +++ b/js/main.js @@ -719,6 +719,7 @@ function loadAssetsAndRenderPageContent() { 'assets/duolingo.yaml', 'assets/eleutherai.yaml', 'assets/ens.yaml', + 'assets/fuse.yaml', 'assets/google.yaml', 'assets/greenbit.yaml', 'assets/hubspot.yaml', @@ -730,9 +731,12 @@ function loadAssetsAndRenderPageContent() { 'assets/juni.yaml', 'assets/kakaobrain.yaml', 'assets/khan.yaml', + 'assets/konan.yaml', 'assets/kotoba.yaml', + 'assets/ktai.yaml', 'assets/laion.yaml', 'assets/latitude.yaml', + 'assets/lg.yaml', 'assets/linkedin.yaml', 'assets/llm360.yaml', 'assets/lmsys.yaml', @@ -742,6 +746,7 @@ function loadAssetsAndRenderPageContent() { 'assets/microsoft.yaml', 'assets/moreh.yaml', 'assets/naver.yaml', + 'assets/ncsoft.yaml', 'assets/neeva.yaml', 'assets/notion.yaml', 'assets/nous.yaml', @@ -769,6 +774,7 @@ function loadAssetsAndRenderPageContent() { 'assets/sciphi.yaml', 'assets/shanghai.yaml', 'assets/shop.yaml', + 'assets/skt.yaml', 'assets/snap.yaml', 'assets/speak.yaml', 'assets/spotify.yaml', @@ -777,6 +783,7 @@ function loadAssetsAndRenderPageContent() { 'assets/stonybrook.yaml', 'assets/tiger.yaml', 'assets/together.yaml', + 'assets/tokyo.yaml', 'assets/trevor.yaml', 'assets/triml.yaml', 'assets/tsinghua.yaml',