From 4370477fc5fc5245ef14b30566e7626113945792 Mon Sep 17 00:00:00 2001 From: jxue16 <105090474+jxue16@users.noreply.github.com> Date: Fri, 8 Dec 2023 22:45:20 -0800 Subject: [PATCH 1/2] weekly updateincluding Gemini --- assets/01ai.yaml | 3 ++- assets/adept.yaml | 7 ++++-- assets/adobe.yaml | 29 ++++++++++++++++-------- assets/amazon.yaml | 6 +++-- assets/anthropic.yaml | 8 +++++-- assets/argilla.yaml | 23 +++++++++++++++++++ assets/baai.yaml | 13 +++++++---- assets/baichuan.yaml | 4 +++- assets/baidu.yaml | 2 +- assets/beitech.yaml | 6 +++-- assets/boston.yaml | 6 +++-- assets/bytedance.yaml | 6 +++-- assets/cohere.yaml | 3 ++- assets/columbia.yaml | 4 +++- assets/continue.yaml | 6 +++-- assets/eleutherai.yaml | 3 ++- assets/epfl.yaml | 23 +++++++++++++++++++ assets/google.yaml | 27 ++++++++++++++++++++++ assets/huggingface.yaml | 12 ++++++---- assets/kunlun.yaml | 10 +++++---- assets/meta.yaml | 3 ++- assets/microsoft.yaml | 25 ++++++++++++++------- assets/mila.yaml | 35 +++++++++++++++++++++-------- assets/mistral.yaml | 3 ++- assets/moonhub.yaml | 3 ++- assets/mosaic.yaml | 6 +++-- assets/ollama.yaml | 25 +++++++++++++++++++++ assets/openai.yaml | 33 +++++++++++++++++---------- assets/openlemur.yaml | 26 ++++++++++++++++----- assets/peking.yaml | 7 ++++-- assets/salesforce.yaml | 11 +++++---- assets/shanghai.yaml | 3 ++- assets/soochow.yaml | 3 ++- assets/stability.yaml | 4 +++- assets/together.yaml | 50 +++++++++++++++++++++++++++++++++++++++-- assets/toronto.yaml | 9 +++++--- assets/tsinghua.yaml | 1 - assets/twelvelabs.yaml | 2 +- assets/uae.yaml | 21 +++++++++++------ assets/xai.yaml | 10 ++++++--- assets/xwin.yaml | 3 ++- js/main.js | 3 +++ 42 files changed, 380 insertions(+), 107 deletions(-) create mode 100644 assets/argilla.yaml create mode 100644 assets/epfl.yaml create mode 100644 assets/ollama.yaml diff --git a/assets/01ai.yaml b/assets/01ai.yaml index 6432e080..0de18f26 100644 --- a/assets/01ai.yaml +++ b/assets/01ai.yaml @@ -7,7 +7,8 @@ url: https://github.com/01-ai/Yi model_card: https://huggingface.co/01-ai/Yi-34B modality: text; text - analysis: Evaluated on common sense reasoning and reading comprehension, analogous to LLaMA 2's analysis. + analysis: Evaluated on common sense reasoning and reading comprehension, analogous + to LLaMA 2's analysis. size: 34B parameters (dense) dependencies: [] training_emissions: unknown diff --git a/assets/adept.yaml b/assets/adept.yaml index 1684962e..0adefae2 100644 --- a/assets/adept.yaml +++ b/assets/adept.yaml @@ -50,7 +50,8 @@ - type: model name: Fuyu organization: Adept - description: Fuyu is a small version of the multimodal model that powers Adept's core product. + description: Fuyu is a small version of the multimodal model that powers Adept's + core product. created_date: 2023-10-17 url: https://www.adept.ai/blog/fuyu-8b model_card: https://huggingface.co/adept/fuyu-8b @@ -65,6 +66,8 @@ access: open license: CC BY NC 4.0 intended_uses: The model is intended for research purposes only. - prohibited_uses: The model was not trained to be factual or true representations of people or events, and therefore using the model to generate such content is out-of-scope for the abilities of this model. + prohibited_uses: The model was not trained to be factual or true representations + of people or events, and therefore using the model to generate such content + is out-of-scope for the abilities of this model. monitoring: '' feedback: https://huggingface.co/adept/fuyu-8b/discussions diff --git a/assets/adobe.yaml b/assets/adobe.yaml index 9d8b0a44..1a8331af 100644 --- a/assets/adobe.yaml +++ b/assets/adobe.yaml @@ -2,7 +2,10 @@ - type: model name: Firefly Image 2 organization: Adobe - description: Firefly Image 2 is the next generation of generative AI for imaging, bringing significant advancements to creative control and quality, including new Text to Image capabilities now available in the popular Firefly web app where 90% of users are new to Adobe products. + description: Firefly Image 2 is the next generation of generative AI for imaging, + bringing significant advancements to creative control and quality, including + new Text to Image capabilities now available in the popular Firefly web app + where 90% of users are new to Adobe products. created_date: 2023-10-10 url: https://firefly.adobe.com/ model_card: none @@ -17,14 +20,17 @@ access: closed license: unknown intended_uses: creative generation of digital art and images - prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential content. + prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential + content. monitoring: '' feedback: '' - type: model name: Firefly Vector organization: Adobe - description: Firefly Vector is the world’s first generative AI focused on producing vector graphics, bringing Adobe's vector graphic and generative AI expertise directly into Adobe Illustrator workflows with Text to Vector Graphic. + description: Firefly Vector is the world’s first generative AI focused on producing + vector graphics, bringing Adobe's vector graphic and generative AI expertise + directly into Adobe Illustrator workflows with Text to Vector Graphic. created_date: 2023-10-10 url: https://firefly.adobe.com/ model_card: none @@ -39,14 +45,16 @@ access: closed license: unknown intended_uses: creative generation of digital art and images - prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential content. + prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential + content. monitoring: '' feedback: '' - type: model name: Firefly Design organization: Adobe - description: Firefly Design powers instant generation of amazing quality template designs in Adobe Express with the new Text to Template capability. + description: Firefly Design powers instant generation of amazing quality template + designs in Adobe Express with the new Text to Template capability. created_date: 2023-10-10 url: https://firefly.adobe.com/ model_card: none @@ -61,14 +69,17 @@ access: closed license: unknown intended_uses: creative generation of digital art and images - prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential content. + prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential + content. monitoring: '' feedback: '' - type: application name: Firefly organization: Adobe - description: Adobe Firefly is a standalone web application. It offers new ways to ideate, create, and communicate while significantly improving creative workflows using generative AI. + description: Adobe Firefly is a standalone web application. It offers new ways + to ideate, create, and communicate while significantly improving creative workflows + using generative AI. created_date: 2023-03-21 url: https://firefly.adobe.com/ dependencies: [Firefly Image 2, Firefly Vector, Firefly Design] @@ -79,10 +90,10 @@ license: unknown terms_of_service: https://www.adobe.com/legal/licenses-terms/adobe-gen-ai-user-guidelines.html intended_uses: creative generation of digital art and images - prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential content. + prohibited_uses: AI/ML training, attempting to create abusive, illegal, or confidential + content. monitoring: '' feedback: '' monthly_active_users: unknown user_distribution: unknown failures: unknown - diff --git a/assets/amazon.yaml b/assets/amazon.yaml index 5a4334a3..4e634935 100644 --- a/assets/amazon.yaml +++ b/assets/amazon.yaml @@ -32,12 +32,14 @@ - type: model name: FalconLite2 organization: Amazon - description: FalconLite2 is a fine-tuned and quantized Falcon language model, capable of processing long (up to 24K tokens) input sequences. + description: FalconLite2 is a fine-tuned and quantized Falcon language model, + capable of processing long (up to 24K tokens) input sequences. created_date: 2023-08-08 url: https://huggingface.co/amazon/FalconLite2 model_card: https://huggingface.co/amazon/FalconLite2 modality: text; text - analysis: Evaluated against benchmarks that are specifically designed to assess the capabilities of LLMs in handling longer contexts. + analysis: Evaluated against benchmarks that are specifically designed to assess + the capabilities of LLMs in handling longer contexts. size: 40B parameters (dense) dependencies: [Falcon] training_emissions: unknown diff --git a/assets/anthropic.yaml b/assets/anthropic.yaml index c78bba82..62f520b5 100644 --- a/assets/anthropic.yaml +++ b/assets/anthropic.yaml @@ -527,12 +527,16 @@ - type: model name: Claude 2.1 organization: Anthropic - description: Claude 2.1 is an updated version of Claude 2, with an increased context window, less hallucination and tool use. + description: Claude 2.1 is an updated version of Claude 2, with an increased context + window, less hallucination and tool use. created_date: 2023-11-21 url: https://www.anthropic.com/index/claude-2-1 model_card: none modality: text; text - analysis: Evaluated on open-ended conversation accuracy and long context question answering. In evaluations, Claude 2.1 demonstrated a 30% reduction in incorrect answers and a 3-4x lower rate of mistakenly concluding a document supports a particular claim. + analysis: Evaluated on open-ended conversation accuracy and long context question + answering. In evaluations, Claude 2.1 demonstrated a 30% reduction in incorrect + answers and a 3-4x lower rate of mistakenly concluding a document supports a + particular claim. size: unknown dependencies: [] training_emissions: unknown diff --git a/assets/argilla.yaml b/assets/argilla.yaml new file mode 100644 index 00000000..732e4a96 --- /dev/null +++ b/assets/argilla.yaml @@ -0,0 +1,23 @@ +--- +- type: model + name: Notus + organization: 01.AI + description: Notus is an open source LLM, fine-tuned using Direct Preference Optimization + (DPO) and AIF (AI Feedback) techniques. + created_date: 2023-12-01 + url: https://argilla.io/blog/notus7b/ + model_card: https://huggingface.co/argilla/notus-7b-v1 + modality: text; text + analysis: Evaluated on MT-Bench and AlphaEval benchmarks. + size: 7B parameters (dense) + dependencies: [UltraFeedback, Zephyr] + training_emissions: unknown + training_time: unknown + training_hardware: 8 x A100 40GB GPUs + quality_control: '' + access: open + license: MIT + intended_uses: Academic research and free commercial usage + prohibited_uses: '' + monitoring: none + feedback: https://huggingface.co/argilla/notus-7b-v1/discussions diff --git a/assets/baai.yaml b/assets/baai.yaml index 1e7f0d51..ebeabd05 100644 --- a/assets/baai.yaml +++ b/assets/baai.yaml @@ -48,7 +48,8 @@ - type: model name: JudgeLM organization: Beijing Academy of Artificial Intelligence - description: JudgeLM is a fine-tuned to be a scalable judge to evaluate LLMs efficiently and effectively in open-ended benchmarks. + description: JudgeLM is a fine-tuned to be a scalable judge to evaluate LLMs efficiently + and effectively in open-ended benchmarks. created_date: 2023-10-26 url: https://arxiv.org/pdf/2310.17631.pdf model_card: https://huggingface.co/BAAI/JudgeLM-13B-v1.0 @@ -62,14 +63,17 @@ quality_control: '' access: open license: custom non-commercial license - intended_uses: Research on evaluating the performance of large language models and chatbots. + intended_uses: Research on evaluating the performance of large language models + and chatbots. prohibited_uses: none monitoring: none feedback: https://huggingface.co/BAAI/JudgeLM-13B-v1.0/discussions - type: dataset name: JudgeLM Dataset organization: Beijing Academy of Artificial Intelligence - description: JudgeLM Dataset is a novel dataset replete with a rich variety of seed tasks, comprehensive answers from modern LLMs, answers’ grades from the teacher judge, and detailed reasons for judgments. + description: JudgeLM Dataset is a novel dataset replete with a rich variety of + seed tasks, comprehensive answers from modern LLMs, answers’ grades from the + teacher judge, and detailed reasons for judgments. created_date: 2023-10-26 url: https://huggingface.co/datasets/BAAI/JudgeLM-100K datasheet: '' @@ -83,7 +87,8 @@ quality_control: '' access: open license: CC BY NC 4.0 - intended_uses: To be used to conduct instruction-tuning for language models and make the language model able to judge open-ended answer pairs. + intended_uses: To be used to conduct instruction-tuning for language models and + make the language model able to judge open-ended answer pairs. prohibited_uses: none monitoring: none feedback: https://huggingface.co/datasets/BAAI/JudgeLM-100K/discussions diff --git a/assets/baichuan.yaml b/assets/baichuan.yaml index cdbf117e..98a0fd4e 100644 --- a/assets/baichuan.yaml +++ b/assets/baichuan.yaml @@ -2,7 +2,9 @@ - type: model name: Baichuan 2 organization: Baichuan Inc. - description: Baichuan 2 is a series of large-scale multilingual language models containing 7 billion and 13 billion parameters, trained from scratch, on 2.6 trillion tokens. + description: Baichuan 2 is a series of large-scale multilingual language models + containing 7 billion and 13 billion parameters, trained from scratch, on 2.6 + trillion tokens. created_date: 2023-09-20 url: https://arxiv.org/pdf/2309.10305.pdf model_card: none diff --git a/assets/baidu.yaml b/assets/baidu.yaml index 0a0037f6..e3fa09a3 100644 --- a/assets/baidu.yaml +++ b/assets/baidu.yaml @@ -85,7 +85,7 @@ organization: Baidu description: ERNIE-4.0 is a multimodal generalist foundation model. created_date: 2023-10-17 - url: + url: explanation: No official statement from Baidu, so news release is provided. value: https://www.prnewswire.com/news-releases/baidu-launches-ernie-4-0-foundation-model-leading-a-new-wave-of-ai-native-applications-301958681.html model_card: none diff --git a/assets/beitech.yaml b/assets/beitech.yaml index 4d388715..99be7628 100644 --- a/assets/beitech.yaml +++ b/assets/beitech.yaml @@ -2,14 +2,16 @@ - type: dataset name: EXMODD organization: Beijing Institute of Technology - description: EXMODD (Explanatory Multimodal Open-Domain Dialogue dataset) is a dataset built off the proposed MDCF (Multimodal Data Construction Framework). + description: EXMODD (Explanatory Multimodal Open-Domain Dialogue dataset) is a + dataset built off the proposed MDCF (Multimodal Data Construction Framework). created_date: 2023-10-17 url: https://arxiv.org/pdf/2310.10967.pdf datasheet: none modality: image, text size: unknown sample: [] - analysis: Models fine-tuned on EXMODD and earlier dataset Image-Chat and then evaluated on Image-Chat validation set. + analysis: Models fine-tuned on EXMODD and earlier dataset Image-Chat and then + evaluated on Image-Chat validation set. dependencies: [YFCC100M, Image-Chat] included: '' excluded: '' diff --git a/assets/boston.yaml b/assets/boston.yaml index adec6753..12061863 100644 --- a/assets/boston.yaml +++ b/assets/boston.yaml @@ -25,12 +25,14 @@ - type: model name: UFOGen organization: Boston University - description: UFOGen is a novel generative model designed for ultra-fast, one-step text-to-image synthesis. + description: UFOGen is a novel generative model designed for ultra-fast, one-step + text-to-image synthesis. created_date: 2023-11-14 url: https://arxiv.org/pdf/2311.09257.pdf model_card: none modality: text; image - analysis: UFOGen is evaluated on standard image benchmarks against other models fine-tuned with Stable Diffusion. + analysis: UFOGen is evaluated on standard image benchmarks against other models + fine-tuned with Stable Diffusion. size: 900M parameters (dense) dependencies: [Stable Diffusion] training_emissions: unknown diff --git a/assets/bytedance.yaml b/assets/bytedance.yaml index ba8a4cf3..e99c236e 100644 --- a/assets/bytedance.yaml +++ b/assets/bytedance.yaml @@ -2,12 +2,14 @@ - type: model name: SALMONN organization: ByteDance, Tsinghua University - description: SALMONN is a large language model (LLM) enabling speech, audio event, and music inputs. + description: SALMONN is a large language model (LLM) enabling speech, audio event, + and music inputs. created_date: 2023-10-20 url: https://github.com/bytedance/SALMONN model_card: https://huggingface.co/MSIIP/SALMONN modality: audio, text; text - analysis: Evaluated on benchmarks pertaining to speech, music, and other audio recognition. + analysis: Evaluated on benchmarks pertaining to speech, music, and other audio + recognition. size: unknown dependencies: [Whisper, BEATs, Vicuna] training_emissions: unknown diff --git a/assets/cohere.yaml b/assets/cohere.yaml index d7577fb6..43d175a9 100644 --- a/assets/cohere.yaml +++ b/assets/cohere.yaml @@ -452,7 +452,8 @@ - type: model name: Cohere Embedv3 (English) organization: Cohere - description: As of release, Cohere Embedv3 is Cohere's latest and most advanced embeddings model. + description: As of release, Cohere Embedv3 is Cohere's latest and most advanced + embeddings model. created_date: 2023-11-02 url: https://txt.cohere.com/introducing-embed-v3/ model_card: https://huggingface.co/Cohere/Cohere-embed-english-v3.0 diff --git a/assets/columbia.yaml b/assets/columbia.yaml index 58c6c299..b29b354b 100644 --- a/assets/columbia.yaml +++ b/assets/columbia.yaml @@ -23,7 +23,9 @@ - type: model name: Ferret organization: Columbia - description: Ferret is a Multimodal Large Language Model (MLLM) capable of understanding spatial referring of any shape or granularity within an image and accurately grounding open-vocabulary descriptions. + description: Ferret is a Multimodal Large Language Model (MLLM) capable of understanding + spatial referring of any shape or granularity within an image and accurately + grounding open-vocabulary descriptions. created_date: 2023-10-11 url: https://arxiv.org/pdf/2310.07704.pdf model_card: none diff --git a/assets/continue.yaml b/assets/continue.yaml index ea63e2f0..e4a3d9a3 100644 --- a/assets/continue.yaml +++ b/assets/continue.yaml @@ -2,8 +2,10 @@ - type: application name: Continue organization: Continue Dev, Inc. - description: Continue is the open-source autopilot for software development. It is an IDE extension that brings the power of - ChatGPT to VS Code and JetBrains. It’s built to be deeply customizable and continuously learn from development data. + description: Continue is the open-source autopilot for software development. It + is an IDE extension that brings the power of ChatGPT to VS Code and JetBrains. + It’s built to be deeply customizable and continuously learn from development + data. created_date: 2023-07-26 url: https://continue.dev dependencies: diff --git a/assets/eleutherai.yaml b/assets/eleutherai.yaml index 10b7d8ec..5e8072c7 100644 --- a/assets/eleutherai.yaml +++ b/assets/eleutherai.yaml @@ -261,7 +261,8 @@ url: https://arxiv.org/pdf/2310.10631.pdf model_card: https://huggingface.co/EleutherAI/llemma_34b modality: text; text - analysis: Evaluated on math benchmarks in comparison to general large language models. + analysis: Evaluated on math benchmarks in comparison to general large language + models. size: 34B parameters (dense) dependencies: [Proof Pile 2, Code LLaMA] training_emissions: unknown diff --git a/assets/epfl.yaml b/assets/epfl.yaml new file mode 100644 index 00000000..83f19be6 --- /dev/null +++ b/assets/epfl.yaml @@ -0,0 +1,23 @@ +--- +- type: model + name: MediTron + organization: EPFL, Idiap Research Institute, Open Assistant, Yale + description: Meditron is a large-scale medical LLM that remains open-source. + created_date: 2023-11-27 + url: https://arxiv.org/pdf/2311.16079.pdf + model_card: https://huggingface.co/epfl-llm/meditron-70b + modality: text; text + analysis: Evaluated on TruthfulQA as main evaluation benchmark. + size: 70B parameters (dense) + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: open + license: Apache 2.0 + intended_uses: Medical exam question answering, supporting differential diagnosis, + disease information. + prohibited_uses: '' + monitoring: none + feedback: https://huggingface.co/01-ai/Yi-34B/discussions diff --git a/assets/google.yaml b/assets/google.yaml index 521351ba..f8836720 100644 --- a/assets/google.yaml +++ b/assets/google.yaml @@ -1678,3 +1678,30 @@ within specific downstream applications without prior assessment monitoring: Google internal monitoring feedback: Specific queries provided by annotators +- type: model + name: Gemini + organization: Google + description: As of release, Gemini is Google's most capable and flexible AI model, + proficient in multimodal domains. + created_date: 2023-12-06 + url: https://deepmind.google/technologies/gemini/#introduction + model_card: none + modality: text; image, text, video + analysis: Evaluated on standard general, reasoning, math, coding, and multimodal + benchmarks with results that surpass GPT-4 on almost all. + size: + explanation: Comes in sizes Ultra, Pro, and Nano. + value: unknown + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: closed + license: Google + intended_uses: general use large language model that can be used for language, + reasoning, and code tasks. + prohibited_uses: becoming part of a general-purpose service or product or use + within specific downstream applications without prior assessment + monitoring: Google internal monitoring + feedback: none diff --git a/assets/huggingface.yaml b/assets/huggingface.yaml index 7bbf3233..c5594e33 100644 --- a/assets/huggingface.yaml +++ b/assets/huggingface.yaml @@ -27,7 +27,8 @@ - type: model name: Zephyr organization: HuggingFace - description: Zephyr is a series of language models that are trained to act as helpful assistants. + description: Zephyr is a series of language models that are trained to act as + helpful assistants. created_date: 2023-10-11 url: https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha model_card: https://huggingface.co/HuggingFaceH4/zephyr-7b-alpha @@ -71,18 +72,21 @@ - type: dataset name: OBELICS organization: HuggingFace - description: OBELICS is a dataset consisting of 141 million interleaved image-text documents scraped from the web and contains 353 million images. + description: OBELICS is a dataset consisting of 141 million interleaved image-text + documents scraped from the web and contains 353 million images. created_date: 2023-08-22 url: https://huggingface.co/blog/idefics datasheet: https://huggingface.co/datasets/HuggingFaceM4/OBELICS modality: image, text size: 115B tokens sample: [] - analysis: Subset of training dataset evaluated for bias using Data Measurements Tool. + analysis: Subset of training dataset evaluated for bias using Data Measurements + Tool. dependencies: [] included: '' excluded: All images for which creators explicitly requested opt-out of AI training. - quality_control: Sexual and violent content still present in OBELICS even after filtering. + quality_control: Sexual and violent content still present in OBELICS even after + filtering. access: open license: CC-BY-4.0 intended_uses: '' diff --git a/assets/kunlun.yaml b/assets/kunlun.yaml index bccc6696..48aefec9 100644 --- a/assets/kunlun.yaml +++ b/assets/kunlun.yaml @@ -1,13 +1,16 @@ --- - type: model name: Skywork - organization: Kunlun Inc. - description: The Skywork series is a family of large language models (LLMs) trained on a corpus of over 3.2 trillion tokens drawn from both English and Chinese texts. + organization: Kunlun Inc. + description: The Skywork series is a family of large language models (LLMs) trained + on a corpus of over 3.2 trillion tokens drawn from both English and Chinese + texts. created_date: 2023-10-30 url: https://arxiv.org/pdf/2310.19341.pdf model_card: https://huggingface.co/Skywork/Skywork-13B-base modality: text; text - analysis: Evaluated on several popular benchmarks and performance in different fields. + analysis: Evaluated on several popular benchmarks and performance in different + fields. size: 13B parameters (dense) dependencies: [SkyPile] training_emissions: unknown @@ -22,4 +25,3 @@ prohibited_uses: '' monitoring: none feedback: https://huggingface.co/Skywork/Skywork-13B-base/discussions - diff --git a/assets/meta.yaml b/assets/meta.yaml index 203f8af9..1bdfabe2 100644 --- a/assets/meta.yaml +++ b/assets/meta.yaml @@ -747,7 +747,8 @@ - type: model name: MetaCLIP organization: Meta - description: MetaCLIP is a more transparent rendition of CLIP that aims to reveal CLIP's training data curation methods. + description: MetaCLIP is a more transparent rendition of CLIP that aims to reveal + CLIP's training data curation methods. created_date: 2023-10-02 url: https://arxiv.org/pdf/2103.00020.pdf model_card: https://huggingface.co/facebook/metaclip-b32-400m diff --git a/assets/microsoft.yaml b/assets/microsoft.yaml index 9a57b5fc..55a379f7 100644 --- a/assets/microsoft.yaml +++ b/assets/microsoft.yaml @@ -766,7 +766,8 @@ - type: model name: Florence-2 organization: Microsoft - description: WizardCoder empowers Code LLMs with complex instruction fine-tuning, by adapting the Evol-Instruct method to the domain of code. + description: WizardCoder empowers Code LLMs with complex instruction fine-tuning, + by adapting the Evol-Instruct method to the domain of code. created_date: 2023-11-10 url: https://arxiv.org/pdf/2311.06242.pdf model_card: none @@ -794,7 +795,8 @@ modality: image, text size: 1.3B image-text annotations sample: [] - analysis: FLD-5B evaluated in comparison to datasets that power other large-scale image models on standard image benchmarks. + analysis: FLD-5B evaluated in comparison to datasets that power other large-scale + image models on standard image benchmarks. dependencies: [] included: '' excluded: '' @@ -808,7 +810,11 @@ - type: dataset name: OpenOrca organization: Microsoft - description: The OpenOrca dataset is a collection of augmented FLAN Collection data. Currently ~1M GPT-4 completions, and ~3.2M GPT-3.5 completions. It is tabularized in alignment with the distributions presented in the ORCA paper and currently represents a partial completion of the full intended dataset, with ongoing generation to expand its scope. + description: The OpenOrca dataset is a collection of augmented FLAN Collection + data. Currently ~1M GPT-4 completions, and ~3.2M GPT-3.5 completions. It is + tabularized in alignment with the distributions presented in the ORCA paper + and currently represents a partial completion of the full intended dataset, + with ongoing generation to expand its scope. created_date: 2023-06-05 url: https://huggingface.co/datasets/Open-Orca/OpenOrca datasheet: https://huggingface.co/datasets/Open-Orca/OpenOrca @@ -829,7 +835,8 @@ - type: model name: LlongOrca organization: Microsoft - description: LlongOrca is an attempt to make OpenOrca able to function in a Llong context. + description: LlongOrca is an attempt to make OpenOrca able to function in a Llong + context. created_date: 2023-08-01 url: https://huggingface.co/Open-Orca/LlongOrca-7B-16k model_card: https://huggingface.co/Open-Orca/LlongOrca-7B-16k @@ -839,7 +846,7 @@ dependencies: [OpenOrca, LLongMA-2] training_emissions: unknown training_time: 37 hours - training_hardware: 8x A6000-48GB (first-gen) GPUs + training_hardware: 8x A6000-48GB (first-gen) GPUs quality_control: '' access: open license: LLaMA2 @@ -855,7 +862,8 @@ url: https://arxiv.org/pdf/2309.05463.pdf model_card: https://huggingface.co/microsoft/phi-1_5 modality: text; text - analysis: Evaluated on common sense reasoning, language understanding, and multi-step reasoning compared to other SOTA language models. + analysis: Evaluated on common sense reasoning, language understanding, and multi-step + reasoning compared to other SOTA language models. size: 1.3B parameters (dense) dependencies: [phi-1] training_emissions: unknown @@ -863,10 +871,11 @@ training_hardware: 32 A100-40G GPUs quality_control: generic web-crawl data is removed from dataset. access: open - license: + license: explanation: can be found via the license tab at top of https://huggingface.co/microsoft/phi-1_5 value: microsoft research license - intended_uses: Phi-1.5 is best suited for answering prompts using the QA format, the chat format, and the code format. + intended_uses: Phi-1.5 is best suited for answering prompts using the QA format, + the chat format, and the code format. prohibited_uses: '' monitoring: none feedback: https://huggingface.co/microsoft/phi-1_5/discussions diff --git a/assets/mila.yaml b/assets/mila.yaml index ed367edb..f8de5899 100644 --- a/assets/mila.yaml +++ b/assets/mila.yaml @@ -2,63 +2,80 @@ - type: dataset name: ToyMix organization: Mila - Quebec AI Institute - description: ToyMix is the smallest dataset of three extensive and meticulously curated multi-label datasets that cover nearly 100 million molecules and over 3000 sparsely defined tasks. + description: ToyMix is the smallest dataset of three extensive and meticulously + curated multi-label datasets that cover nearly 100 million molecules and over + 3000 sparsely defined tasks. created_date: 2023-10-09 url: https://arxiv.org/pdf/2310.04292.pdf datasheet: none modality: molecules, tasks size: 13B labels of quantum and biological nature. sample: [] - analysis: Models of size 150k parameters trained on ToyMix and compared to models trained on its dependencies across GNN baselines. + analysis: Models of size 150k parameters trained on ToyMix and compared to models + trained on its dependencies across GNN baselines. dependencies: [QM9, TOX21, ZINC12K] included: '' excluded: '' quality_control: '' access: open license: CC BY-NC-SA 4.0 - intended_uses: The datasets are intended to be used in an academic setting for training molecular GNNs with orders of magnitude more parameters than current large models. Further, the ToyMix dataset is intended to be used in a multi-task setting, meaning that a single model should be trained to predict them simultaneously. + intended_uses: The datasets are intended to be used in an academic setting for + training molecular GNNs with orders of magnitude more parameters than current + large models. Further, the ToyMix dataset is intended to be used in a multi-task + setting, meaning that a single model should be trained to predict them simultaneously. prohibited_uses: none monitoring: none feedback: none - type: dataset name: LargeMix organization: Mila - Quebec AI Institute - description: LargeMix is the middle-sized dataset of three extensive and meticulously curated multi-label datasets that cover nearly 100 million molecules and over 3000 sparsely defined tasks. + description: LargeMix is the middle-sized dataset of three extensive and meticulously + curated multi-label datasets that cover nearly 100 million molecules and over + 3000 sparsely defined tasks. created_date: 2023-10-09 url: https://arxiv.org/pdf/2310.04292.pdf datasheet: none modality: molecules, tasks size: 13B labels of quantum and biological nature. sample: [] - analysis: Models of size between 4M and 6M parameters trained for 200 epochs on LargeMix and compared to models trained on its dependencies across GNN baselines. + analysis: Models of size between 4M and 6M parameters trained for 200 epochs on + LargeMix and compared to models trained on its dependencies across GNN baselines. dependencies: [L1000 VCAP, L1000 MCF7, PCBA1328, PCQM4M_G25_N4] included: '' excluded: '' quality_control: '' access: open license: CC BY-NC-SA 4.0 - intended_uses: The datasets are intended to be used in an academic setting for training molecular GNNs with orders of magnitude more parameters than current large models. Further, the LargeMix dataset is intended to be used in a multi-task setting, meaning that a single model should be trained to predict them simultaneously. + intended_uses: The datasets are intended to be used in an academic setting for + training molecular GNNs with orders of magnitude more parameters than current + large models. Further, the LargeMix dataset is intended to be used in a multi-task + setting, meaning that a single model should be trained to predict them simultaneously. prohibited_uses: none monitoring: none feedback: none - type: dataset name: UltraLarge organization: Mila - Quebec AI Institute - description: UltraLarge is the largest dataset of three extensive and meticulously curated multi-label datasets that cover nearly 100 million molecules and over 3000 sparsely defined tasks. + description: UltraLarge is the largest dataset of three extensive and meticulously + curated multi-label datasets that cover nearly 100 million molecules and over + 3000 sparsely defined tasks. created_date: 2023-10-09 url: https://arxiv.org/pdf/2310.04292.pdf datasheet: none modality: molecules, tasks size: 13B labels of quantum and biological nature. sample: [] - analysis: Models of size between 4M and 6M parameters trained for 50 epochs on UltraLarge and compared to models trained on its dependencies across GNN baselines. + analysis: Models of size between 4M and 6M parameters trained for 50 epochs on + UltraLarge and compared to models trained on its dependencies across GNN baselines. dependencies: [PM6_83M] included: '' excluded: '' quality_control: '' access: open license: CC BY-NC-SA 4.0 - intended_uses: The datasets are intended to be used in an academic setting for training molecular GNNs with orders of magnitude more parameters than current large models. + intended_uses: The datasets are intended to be used in an academic setting for + training molecular GNNs with orders of magnitude more parameters than current + large models. prohibited_uses: none monitoring: none feedback: none diff --git a/assets/mistral.yaml b/assets/mistral.yaml index c9afff9f..dae2be37 100644 --- a/assets/mistral.yaml +++ b/assets/mistral.yaml @@ -7,7 +7,8 @@ url: https://mistral.ai/news/announcing-mistral-7b/ model_card: https://huggingface.co/mistralai/Mistral-7B-v0.1 modality: text; text - analysis: Evaluated in comparison to LLaMA series models on standard language benchmarks. + analysis: Evaluated in comparison to LLaMA series models on standard language + benchmarks. size: 7.3B parameters (dense) dependencies: [] training_emissions: unknown diff --git a/assets/moonhub.yaml b/assets/moonhub.yaml index 9ee115f3..82dc8d40 100644 --- a/assets/moonhub.yaml +++ b/assets/moonhub.yaml @@ -2,7 +2,8 @@ - type: application name: Moonhub Recruiter organization: Moonhub - description: Moonhub Recruiter is the world's first AI-powered recruiter providing sourcing and recruiting services for startups and growing businesses. + description: Moonhub Recruiter is the world's first AI-powered recruiter providing + sourcing and recruiting services for startups and growing businesses. created_date: 2023-10-11 url: https://www.moonhub.ai/ dependencies: [Cohere Base] diff --git a/assets/mosaic.yaml b/assets/mosaic.yaml index 7de93626..f66aac72 100644 --- a/assets/mosaic.yaml +++ b/assets/mosaic.yaml @@ -24,7 +24,8 @@ - type: model name: CommonCanvas organization: Cornell University, Mosaic - description: CommonCanvas is a text-to-image model trained solely on Creative Commons licensed images. + description: CommonCanvas is a text-to-image model trained solely on Creative + Commons licensed images. created_date: 2023-10-25 url: https://arxiv.org/pdf/2310.16825.pdf model_card: none @@ -45,7 +46,8 @@ - type: dataset name: CommonCatalog organization: Mosaic - description: CommonCatalog is a curated dataset of CommonCrawl images and synthetic captions. + description: CommonCatalog is a curated dataset of CommonCrawl images and synthetic + captions. created_date: 2023-10-25 url: https://arxiv.org/pdf/2310.16825.pdf datasheet: https://github.com/mosaicml/diffusion/blob/main/assets/common-canvas.md diff --git a/assets/ollama.yaml b/assets/ollama.yaml new file mode 100644 index 00000000..a35d7b17 --- /dev/null +++ b/assets/ollama.yaml @@ -0,0 +1,25 @@ +--- +- type: model + name: Starling + organization: Ollama + description: Starling is a large language model trained by reinforcement learning + from AI feedback focused on improving chatbot helpfulness. + created_date: 2023-11-02 + url: https://starling.cs.berkeley.edu/ + model_card: https://huggingface.co/berkeley-nest/Starling-LM-7B-alpha + modality: text; text + analysis: Mainly evaluated on MT-Bench and AlpacaEval, which are GPT-4-based comparisons. + size: 7B parameters (dense) + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: open + license: + explanation: License can be found at model card link. + value: custom + intended_uses: Academic research and free commercial usage + prohibited_uses: '' + monitoring: none + feedback: https://huggingface.co/01-ai/Yi-34B/discussions diff --git a/assets/openai.yaml b/assets/openai.yaml index f9e66064..5a3fa740 100644 --- a/assets/openai.yaml +++ b/assets/openai.yaml @@ -1355,11 +1355,14 @@ - type: model name: DALL·E 3 organization: OpenAI - description: DALL·E 3 is an artificial intelligence model that takes a text - prompt and/or existing image as an input and generates a new image as an output - The model is now in research preview, and will be available to ChatGPT Plus and Enterprise customers in October. + description: DALL·E 3 is an artificial intelligence model that takes a text prompt + and/or existing image as an input and generates a new image as an output The + model is now in research preview, and will be available to ChatGPT Plus and + Enterprise customers in October. created_date: - explanation: OpenAI announced that DALL·E 3 was coming soon in a tweet on 2023-09-20. Users could begin experimenting with DALL·E 3 in research preview in early October. + explanation: OpenAI announced that DALL·E 3 was coming soon in a tweet on 2023-09-20. + Users could begin experimenting with DALL·E 3 in research preview in early + October. value: 2023-09-20 url: https://openai.com/dall-e-3 model_card: none @@ -1371,19 +1374,25 @@ training_emissions: unknown training_time: unknown training_hardware: unknown - quality_control: DALL·E 3 has mitigations to decline requests that ask for a public figure by name. We improved safety performance in risk areas like generation of public figures and harmful biases related to visual over/under-representation, in partnership with red teamers—domain experts who stress-test the model—to help inform our risk assessment and mitigation efforts in areas like propaganda and misinformation. + quality_control: DALL·E 3 has mitigations to decline requests that ask for a public + figure by name. We improved safety performance in risk areas like generation + of public figures and harmful biases related to visual over/under-representation, + in partnership with red teamers—domain experts who stress-test the model—to + help inform our risk assessment and mitigation efforts in areas like propaganda + and misinformation. access: - explanation: DALL·E 3 is now in research preview, and will be available to ChatGPT Plus and Enterprise customers in October, via the API and in Labs later this fall. + explanation: DALL·E 3 is now in research preview, and will be available to ChatGPT + Plus and Enterprise customers in October, via the API and in Labs later this + fall. value: limited license: unknown - intended_uses: The intended use of the DALL·E 3 Preview at this time is for - personal, non-commercial exploration and research purposes by people who are - interested in understanding the potential uses of these capabilities + intended_uses: The intended use of the DALL·E 3 Preview at this time is for personal, + non-commercial exploration and research purposes by people who are interested + in understanding the potential uses of these capabilities prohibited_uses: Use of the model is governed by the OpenAI Content Policy, which prohibits posting of G rated content. Users are not allowed to utilize the model in commercial products in the preview version. monitoring: Uses of the model are monitored. In the preview version, any user - can flag content. The specific policies for monitoring are not disclosed, - but possible measures include disabling of accounts violating the content + can flag content. The specific policies for monitoring are not disclosed, but + possible measures include disabling of accounts violating the content feedback: Feedback can be provided at openai.com - diff --git a/assets/openlemur.yaml b/assets/openlemur.yaml index 7464f631..2cedfa9f 100644 --- a/assets/openlemur.yaml +++ b/assets/openlemur.yaml @@ -2,14 +2,23 @@ - type: model name: Lemur organization: OpenLemur - description: Lemur is an openly accessible language model optimized for both natural language and coding capabilities to serve as the backbone of versatile language agents. + description: Lemur is an openly accessible language model optimized for both natural + language and coding capabilities to serve as the backbone of versatile language + agents. created_date: 2023-10-10 url: https://arxiv.org/pdf/2310.06830.pdf model_card: https://huggingface.co/OpenLemur/lemur-70b-v1 modality: code, text; code, text analysis: Evaluated on text and code benchmarks in comparison to other models. size: 70B parameters (dense) - dependencies: [LLaMA 2, The Stack, RefinedWeb, RedPajama, Common Crawl, Wikipedia, ArXiv] + dependencies: + - LLaMA 2 + - The Stack + - RefinedWeb + - RedPajama + - Common Crawl + - Wikipedia + - ArXiv training_emissions: unknown training_time: unknown training_hardware: TPUv4-512 pod @@ -23,14 +32,21 @@ - type: model name: Lemur-Chat organization: OpenLemur - description: Lemur-Chat is an openly accessible language model optimized for both natural language and coding capabilities to serve as the backbone of versatile language agents. + description: Lemur-Chat is an openly accessible language model optimized for both + natural language and coding capabilities to serve as the backbone of versatile + language agents. created_date: 2023-10-10 url: https://arxiv.org/pdf/2310.06830.pdf model_card: https://huggingface.co/OpenLemur/lemur-70b-chat-v1 modality: text; text analysis: Evaluated on text and code benchmarks in comparison to other models. size: 70B parameters (dense) - dependencies: [Lemur, OpenAssistant 1, OpenOrca, ShareGPT & ChatLogs, Evol-CodeAlpaca data] + dependencies: + - Lemur + - OpenAssistant 1 + - OpenOrca + - ShareGPT & ChatLogs + - Evol-CodeAlpaca data training_emissions: unknown training_time: unknown training_hardware: unknown @@ -40,4 +56,4 @@ intended_uses: '' prohibited_uses: '' monitoring: '' - feedback: https://huggingface.co/OpenLemur/lemur-70b-chat-v1/discussions \ No newline at end of file + feedback: https://huggingface.co/OpenLemur/lemur-70b-chat-v1/discussions diff --git a/assets/peking.yaml b/assets/peking.yaml index 93883cd4..b0884557 100644 --- a/assets/peking.yaml +++ b/assets/peking.yaml @@ -2,12 +2,15 @@ - type: model name: JARVIS-1 organization: Peking University Institute for Artificial Intelligence - description: JARVIS-1 is an open-world agent that can perceive multimodal input (visual observations and human instructions), generate sophisticated plans, and perform embodied control, all within the popular yet challenging open-world Minecraft universe. + description: JARVIS-1 is an open-world agent that can perceive multimodal input + (visual observations and human instructions), generate sophisticated plans, + and perform embodied control, all within the popular yet challenging open-world + Minecraft universe. created_date: 2023-11-10 url: https://arxiv.org/pdf/2311.05997.pdf model_card: none modality: text; in-game actions - analysis: Compared with other multi-task, instruction-following agents. + analysis: Compared with other multi-task, instruction-following agents. size: unknown dependencies: [] training_emissions: unknown diff --git a/assets/salesforce.yaml b/assets/salesforce.yaml index 71df410c..738edb69 100644 --- a/assets/salesforce.yaml +++ b/assets/salesforce.yaml @@ -99,15 +99,18 @@ - type: model name: BLIP-2 organization: Salesforce - description: BLIP-2 is a model that employs a generic and efficient pretraining strategy that bootstraps vision-language pre-training from off-the-shelf frozen pre-trained image encoders and frozen large language models. + description: BLIP-2 is a model that employs a generic and efficient pretraining + strategy that bootstraps vision-language pre-training from off-the-shelf frozen + pre-trained image encoders and frozen large language models. created_date: 2023-01-30 url: https://arxiv.org/pdf/2301.12597.pdf model_card: https://huggingface.co/Salesforce/blip2-opt-2.7b modality: text; image - analysis: BLIP-2 achieves state-of-the-art performance on various vision-language tasks, despite having significantly fewer trainable parameters than existing methods + analysis: BLIP-2 achieves state-of-the-art performance on various vision-language + tasks, despite having significantly fewer trainable parameters than existing + methods size: 2.7B parameters (dense) - dependencies: - - OPT + dependencies: [OPT] training_emissions: unknown training_time: less than 9 days training_hardware: unknown diff --git a/assets/shanghai.yaml b/assets/shanghai.yaml index e75a75bc..26105e3a 100644 --- a/assets/shanghai.yaml +++ b/assets/shanghai.yaml @@ -76,7 +76,8 @@ - type: model name: InternLM organization: Shanghai AI Laboratory - description: InternLM is a high-quality language model proficient in English, Chinese, and code. + description: InternLM is a high-quality language model proficient in English, + Chinese, and code. created_date: 2023-09-20 url: https://github.com/InternLM/InternLM model_card: https://huggingface.co/internlm/internlm-20b diff --git a/assets/soochow.yaml b/assets/soochow.yaml index fca7cd53..31e0b025 100644 --- a/assets/soochow.yaml +++ b/assets/soochow.yaml @@ -2,7 +2,8 @@ - type: model name: OpenBA organization: Soochow University - description: OpenBA is an open-sourced 15B bilingual (English + Chinese) asymmetric seq2seq model. + description: OpenBA is an open-sourced 15B bilingual (English + Chinese) asymmetric + seq2seq model. created_date: 2023-10-01 url: https://arxiv.org/pdf/2309.10706.pdf model_card: https://huggingface.co/OpenBA/OpenBA-LM diff --git a/assets/stability.yaml b/assets/stability.yaml index 7933c08d..38f86d4f 100644 --- a/assets/stability.yaml +++ b/assets/stability.yaml @@ -97,7 +97,9 @@ - type: application name: Sky Replacer organization: Stability AI - description: Sky Replacer is an exciting new tool that allows users to replace the color and aesthetic of the sky in their original photos with a selection of nine alternatives to improve the overall look and feel of the image. + description: Sky Replacer is an exciting new tool that allows users to replace + the color and aesthetic of the sky in their original photos with a selection + of nine alternatives to improve the overall look and feel of the image. created_date: 2023-11-01 url: https://clipdrop.co/real-estate/sky-replacer dependencies: [] diff --git a/assets/together.yaml b/assets/together.yaml index f1075a54..7da614a8 100644 --- a/assets/together.yaml +++ b/assets/together.yaml @@ -154,7 +154,10 @@ - type: dataset name: RedPajama-Data-v2 organization: Together - description: RedPajama-Data-v2 is a new version of the RedPajama dataset, with 30 trillion filtered and deduplicated tokens (100+ trillions raw) from 84 CommonCrawl dumps covering 5 languages, along with 40+ pre-computed data quality annotations that can be used for further filtering and weighting. + description: RedPajama-Data-v2 is a new version of the RedPajama dataset, with + 30 trillion filtered and deduplicated tokens (100+ trillions raw) from 84 CommonCrawl + dumps covering 5 languages, along with 40+ pre-computed data quality annotations + that can be used for further filtering and weighting. created_date: 2023-10-30 url: https://together.ai/blog/redpajama-data-v2 datasheet: '' @@ -168,7 +171,50 @@ quality_control: tokens filtered and deduplicated access: open license: Apache 2.0 - intended_uses: To be used as the start of a larger, community-driven development of large-scale datasets for LLMs. + intended_uses: To be used as the start of a larger, community-driven development + of large-scale datasets for LLMs. prohibited_uses: '' monitoring: '' feedback: Feedback can be sent to Together via https://www.together.ai/contact +- type: model + name: StripedHyena + organization: Together + description: StripedHyena is an LLM and the first alternative model competitive with the best open-source Transformers in short and long-context evaluations, according to Together. + created_date: 2023-12-08 + url: https://www.together.ai/blog/stripedhyena-7b + model_card: https://huggingface.co/togethercomputer/StripedHyena-Hessian-7B + modality: text; text + analysis: Model evaluated on a suite of short-context task benchmarks. + size: 7B parameters (dense) + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: open + license: Apache 2.0 + intended_uses: '' + prohibited_uses: '' + monitoring: '' + feedback: https://huggingface.co/togethercomputer/StripedHyena-Hessian-7B/discussions +- type: model + name: StripedHyena Nous + organization: Together + description: StripedHyena Nous is an LLM and chatbot, along with the first alternative model competitive with the best open-source Transformers in short and long-context evaluations, according to Together. + created_date: 2023-12-08 + url: https://www.together.ai/blog/stripedhyena-7b + model_card: https://huggingface.co/togethercomputer/StripedHyena-Nous-7B + modality: text; text + analysis: Model evaluated on a suite of short-context task benchmarks. + size: 7B parameters (dense) + dependencies: [] + training_emissions: unknown + training_time: unknown + training_hardware: unknown + quality_control: '' + access: open + license: Apache 2.0 + intended_uses: '' + prohibited_uses: '' + monitoring: '' + feedback: https://huggingface.co/togethercomputer/StripedHyena-Nous-7B/discussions diff --git a/assets/toronto.yaml b/assets/toronto.yaml index ae498064..8d0378ab 100644 --- a/assets/toronto.yaml +++ b/assets/toronto.yaml @@ -2,18 +2,21 @@ - type: dataset name: OpenWebMath organization: University of Toronto - description: OpenWebMath is an open dataset containing 14.7B tokens of mathematical webpages from Common Crawl, inspired by Minerva. + description: OpenWebMath is an open dataset containing 14.7B tokens of mathematical + webpages from Common Crawl, inspired by Minerva. created_date: 2023-10-10 url: https://arxiv.org/pdf/2310.06786.pdf datasheet: Can be found at section E of https://arxiv.org/pdf/2310.06786.pdf modality: text, mathematical tokens size: 14.7B documents sample: [] - analysis: Compared models trained on OpenWebMath for 1 epoch to models trained on The Pile and ProofPile on mathematics benchmarks. + analysis: Compared models trained on OpenWebMath for 1 epoch to models trained + on The Pile and ProofPile on mathematics benchmarks. dependencies: [Common Crawl] included: '' excluded: '' - quality_control: Documents are filtered, processed for mathematical value, deduplicated, and then the largest documents are manually inspected for quality. + quality_control: Documents are filtered, processed for mathematical value, deduplicated, + and then the largest documents are manually inspected for quality. access: open license: ODC-By 1.0 intended_uses: Language model pretraining, finetuning, and evaluation. diff --git a/assets/tsinghua.yaml b/assets/tsinghua.yaml index 1a6b4b9b..9b33bb31 100644 --- a/assets/tsinghua.yaml +++ b/assets/tsinghua.yaml @@ -158,4 +158,3 @@ prohibited_uses: none monitoring: none feedback: none - diff --git a/assets/twelvelabs.yaml b/assets/twelvelabs.yaml index 19fedc6d..0925ae0b 100644 --- a/assets/twelvelabs.yaml +++ b/assets/twelvelabs.yaml @@ -7,7 +7,7 @@ url: https://app.twelvelabs.io/blog/introducing-pegasus-1 model_card: none modality: video; text - analysis: Evaluated in comparison to SOTA video-to-language models. + analysis: Evaluated in comparison to SOTA video-to-language models. size: 80B parameters (dense) dependencies: [MSR-VTT, Video-ChatGPT Video Descriptions Dataset] training_emissions: unknown diff --git a/assets/uae.yaml b/assets/uae.yaml index ecc70ec1..0c5e6998 100644 --- a/assets/uae.yaml +++ b/assets/uae.yaml @@ -5,21 +5,27 @@ description: Falcon-40B is a 40B parameters causal decoder-only model built by TII and trained on 1,000B tokens of RefinedWeb enhanced with curated corpora. created_date: 2023-06-14 - url: https://huggingface.co/tiiuae/falcon-40b + url: https://arxiv.org/pdf/2311.16867.pdf model_card: https://huggingface.co/tiiuae/falcon-40b modality: text; text - analysis: '' + analysis: Evaluated in 1-shot against the PaLM models, with the tasks of the paper + "Language models are few-shot learners" (Brown et al., 2020); (2) on a small + set of few-shot tasks reported by the GPT-4 paper; (3) against state-of-the-art + models across common sense, question answering, and code tasks; (4) against + models which also report results from the EAI Harness, for which we are able + to compare with identical prompts and metrics. size: 40B parameters (dense) dependencies: [RefinedWeb] - training_emissions: '' + training_emissions: unknown training_time: 2 months training_hardware: 384 A100 40GB GPUs quality_control: '' access: open license: Apache 2.0 - intended_uses: '' + intended_uses: Research on large language models; as a foundation for further + specialization for specific use cases. prohibited_uses: irresponsible or harmful use or production use without adequate - assessment of risks and mitigation + assessment of risks and mitigation. monitoring: None feedback: https://huggingface.co/tiiuae/falcon-40b/discussions - type: dataset @@ -52,7 +58,7 @@ description: Falcon-180B is a 180B parameters causal decoder-only model built by TII and trained on 3,500B tokens of RefinedWeb enhanced with curated corpora. created_date: 2023-09-06 - url: https://falconllm.tii.ae/falcon-models.html + url: https://arxiv.org/pdf/2311.16867.pdf model_card: https://huggingface.co/tiiuae/falcon-180B modality: text; text analysis: Falcon-180B outperforms LLaMA-2, StableLM, RedPajama, MPT on the Open @@ -65,7 +71,8 @@ quality_control: '' access: open license: unknown - intended_uses: '' + intended_uses: Research on large language models; as a foundation for further + specialization for specific use cases. prohibited_uses: Production use without adequate assessment of risks and mitigation; any use cases which may be considered irresponsible or harmful. monitoring: None diff --git a/assets/xai.yaml b/assets/xai.yaml index 6470e518..3b82ea52 100644 --- a/assets/xai.yaml +++ b/assets/xai.yaml @@ -2,12 +2,14 @@ - type: model name: Grok-1 organization: xAI - description: Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, intended to answer almost anything and even suggest what questions to ask. + description: Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, + intended to answer almost anything and even suggest what questions to ask. created_date: 2023-11-04 url: https://x.ai/ model_card: https://x.ai/model-card/ modality: text; text - analysis: Grok-1 was evaluated on a range of reasoning benchmark tasks and on curated foreign mathematic examination questions. + analysis: Grok-1 was evaluated on a range of reasoning benchmark tasks and on + curated foreign mathematic examination questions. size: unknown dependencies: [] training_emissions: unknown @@ -16,7 +18,9 @@ quality_control: none access: closed license: unknown - intended_uses: Grok-1 is intended to be used as the engine behind Grok for natural language processing tasks including question answering, information retrieval, creative writing and coding assistance. + intended_uses: Grok-1 is intended to be used as the engine behind Grok for natural + language processing tasks including question answering, information retrieval, + creative writing and coding assistance. prohibited_uses: none monitoring: unknown feedback: none diff --git a/assets/xwin.yaml b/assets/xwin.yaml index a421d71e..d8a58b21 100644 --- a/assets/xwin.yaml +++ b/assets/xwin.yaml @@ -2,7 +2,8 @@ - type: model name: Xwin-LM organization: Xwin - description: Xwin-LM is a LLM, which on release, ranked top 1 on AlpacaEval, becoming the first to surpass GPT-4 on this benchmark. + description: Xwin-LM is a LLM, which on release, ranked top 1 on AlpacaEval, becoming + the first to surpass GPT-4 on this benchmark. created_date: 2023-09-20 url: https://huggingface.co/Xwin-LM/Xwin-LM-70B-V0.1 model_card: https://huggingface.co/Xwin-LM/Xwin-LM-70B-V0.1 diff --git a/js/main.js b/js/main.js index 5049ac34..1db28eaf 100644 --- a/js/main.js +++ b/js/main.js @@ -689,6 +689,9 @@ function loadAssetsAndRenderPageContent() { 'assets/cmu.yaml', 'assets/cohere.yaml', 'assets/databricks.yaml', + 'assets/ollama.yaml', + 'assets/argilla.yaml', + 'assets/epfl.yaml', 'assets/deepmind.yaml', 'assets/duckduckgo.yaml', 'assets/duolingo.yaml', From 68d6596111c3a0630e68a206476ce077c47ba3f0 Mon Sep 17 00:00:00 2001 From: jxue16 <105090474+jxue16@users.noreply.github.com> Date: Sat, 9 Dec 2023 13:30:24 -0800 Subject: [PATCH 2/2] resolve changes --- assets/argilla.yaml | 2 +- assets/epfl.yaml | 4 ++-- assets/google.yaml | 4 ++-- assets/ollama.yaml | 6 ++---- assets/together.yaml | 4 ++-- 5 files changed, 9 insertions(+), 11 deletions(-) diff --git a/assets/argilla.yaml b/assets/argilla.yaml index 732e4a96..ee436dba 100644 --- a/assets/argilla.yaml +++ b/assets/argilla.yaml @@ -1,7 +1,7 @@ --- - type: model name: Notus - organization: 01.AI + organization: Argilla description: Notus is an open source LLM, fine-tuned using Direct Preference Optimization (DPO) and AIF (AI Feedback) techniques. created_date: 2023-12-01 diff --git a/assets/epfl.yaml b/assets/epfl.yaml index 83f19be6..87733460 100644 --- a/assets/epfl.yaml +++ b/assets/epfl.yaml @@ -15,9 +15,9 @@ training_hardware: unknown quality_control: '' access: open - license: Apache 2.0 + license: LLaMA 2 intended_uses: Medical exam question answering, supporting differential diagnosis, disease information. prohibited_uses: '' monitoring: none - feedback: https://huggingface.co/01-ai/Yi-34B/discussions + feedback: https://huggingface.co/epfl-llm/meditron-7b/discussions diff --git a/assets/google.yaml b/assets/google.yaml index f8836720..1bedd680 100644 --- a/assets/google.yaml +++ b/assets/google.yaml @@ -1671,7 +1671,7 @@ quality_control: Employed de-duplication, removal of sensitive-PII and filtering. Added control tokens marking toxicity of text. access: open - license: Google + license: unknown intended_uses: general use large language model that can be used for language, reasoning, and code tasks. prohibited_uses: becoming part of a general-purpose service or product or use @@ -1698,7 +1698,7 @@ training_hardware: unknown quality_control: '' access: closed - license: Google + license: unknown intended_uses: general use large language model that can be used for language, reasoning, and code tasks. prohibited_uses: becoming part of a general-purpose service or product or use diff --git a/assets/ollama.yaml b/assets/ollama.yaml index a35d7b17..9274e846 100644 --- a/assets/ollama.yaml +++ b/assets/ollama.yaml @@ -16,10 +16,8 @@ training_hardware: unknown quality_control: '' access: open - license: - explanation: License can be found at model card link. - value: custom + license: CC BY NC 4.0 intended_uses: Academic research and free commercial usage prohibited_uses: '' monitoring: none - feedback: https://huggingface.co/01-ai/Yi-34B/discussions + feedback: https://huggingface.co/berkeley-nest/Starling-LM-7B-alpha/discussions diff --git a/assets/together.yaml b/assets/together.yaml index 7da614a8..5bc511d0 100644 --- a/assets/together.yaml +++ b/assets/together.yaml @@ -186,7 +186,7 @@ modality: text; text analysis: Model evaluated on a suite of short-context task benchmarks. size: 7B parameters (dense) - dependencies: [] + dependencies: [Hyena, RedPajama-Data] training_emissions: unknown training_time: unknown training_hardware: unknown @@ -207,7 +207,7 @@ modality: text; text analysis: Model evaluated on a suite of short-context task benchmarks. size: 7B parameters (dense) - dependencies: [] + dependencies: [Hyena, RedPajama-Data] training_emissions: unknown training_time: unknown training_hardware: unknown