A model fine-tuned to suggest code completion based on the context in code that's written.

This document appears to be related to a software or technology product called "code-gecko". It includes information such as its licensing, release date, capabilities, and maximum context. However, no description of the product is provided.

PaLM-E is a powerful, multimodal language model designed for a variety of NLP tasks. With its advanced capabilities and cutting-edge technology, it's a top choice for researchers and developers alike.

PaLM-E

Researchers from Google's Robotics team recently announced PaLM-E, a combination of their PaLM and Vision Transformer (ViT) models designed for controlling robots. PaLM-E handles multimodal input data from robotic sensors and outputs text commands to control the robot's actuators.

This document describes a product called "palm-e," including its release date, capabilities, and deprecation status.

Flan-T5/PaLM is a language model for scaling instructions, fine-tuned for optimal performance. With its powerful capabilities and comprehensive training data, this model is an excellent choice for natural language processing tasks. It is available under a permissive license and is highly affordable.

Flan-T5/PaLM

Scaling Instruction-Finetuned Language Models

Flan-T5/PaLM is a language model designed for natural language processing tasks, available under a permissive license and highly affordable.

UL2 is a language learning platform that unifies paradigms. It offers training data, comes in a variety of sizes, and is available under a flexible license. With competitive pricing, UL2 is a versatile tool that can help users achieve their language learning goals.

UL2 is a unified framework for pretraining models that are universally effective across datasets and setups. UL2 uses Mixture-of-Denoisers (MoD), apre-training objective that combines diverse pre-training paradigms together. UL2 introduces a notion of mode switching, wherein downstream fine-tuning is associated with specific pre-training schemes.

This document appears to be a template or metadata file for an open source project called "ul-2," with release information and links to announcements.

PaLM is a language modeling tool that uses pathways to scale up. It has a flexible license, and offers a range of capabilities for natural language processing.

Palm

PaLM: Scaling Language Modeling with Pathways

This document describes a software product called PALM. It is a closed-source language model that was released on April 4, 2022. More information can be found in the announcement link provided.

Minerva is a closed-source text model built to foster quantitative reasoning. It leverages Pathways Language Model (PaLM) and is trained on a large dataset of scientific papers and web pages containing mathematical expressions. It was released on 30/06/2022 and is capable of solving complex mathematical problems.

Minerva

To promote quantitative reasoning, Minerva builds on the Pathways Language Model (PaLM), with further training on a 118GB dataset of scientific papers from the arXiv preprint server and web pages that contain mathematical expressions using LaTeX, MathJax, or other mathematical typesetting formats.

Minerva is a closed-source text model designed to promote quantitative reasoning by solving complex mathematical problems.

LaMDA is a language model designed for dialog applications. It uses cutting-edge natural language processing techniques to facilitate more natural and engaging conversations. With its advanced capabilities and versatile applications, LaMDA represents the future of conversational AI.

Lamda

LaMDA: Language Models for Dialog Applications

This document contains information about the closed source AI language model called "Lamda".

Flan is a text-based tool that utilizes finetuned language models as zero-shot learners. It is a powerful tool for natural language processing and can be used for a variety of tasks. Flan is available under a non-deprecated license and has no pricing data.

Flan

Finetuned Language Models are Zero-Shot Learners

Flan is a text-based tool that utilizes finetuned language models as zero-shot learners.

T5 is a text-based model that explores the limits of transfer learning using a unified transformer architecture. It is designed to handle a wide range of natural language processing tasks, and is available for use under a non-deprecated license.

Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

This document describes a software project called t5, which is open source and consists of text. It was released on 24/02/2020 and is not deprecated.

Codey for Code Generation is a closed source text model that supports generating code based on natural language descriptions. It can create functions, web pages, and unit tests in Go, GoogleSQL, Java, JavaScript, Python, and TypeScript. Supported infrastructure as code interfaces include Google Cloud CLI, Kubernetes Resource Model (KRM), and Terraform. Expected release date is 07/04/2023.

Codey for Code Generation

code-bison is the name of the model that supports code generation. It's a foundation model that generates code based on a natural language description. The type of content that code-bison can create includes functions, web pages, and unit tests. code-bison is supported by the code generation Codey API. The Codey APIs are in the PaLM API family.

Supported coding languages
The Codey APIs support the following programming languages:
• Go
• GoogleSQL
• Java
• JavaScript
• Python
• TypeScript
Supported code infrastructure interfaces
The Codey API support the following infrastructure as code interfaces:
• Google Cloud CLI
• Kubernetes Resource Model (KRM)
• Terraform

This document describes the Code Bison project, which is a closed-source text-based AI model that will be released on June 29, 2023. It will be available on the Google Cloud Platform's Vertex AI and will have the capability to analyze code with a maximum context of 6.

Codey for Code Chat is a closed source chat API model designed for developers seeking code-related assistance. It supports multi-turn conversations specialized for code and can be accessed through the Codey APIs. Ideal for back-and-forth interactions, Codey for Code Chat offers continuous conversation for code tasks. This model has a release date of 07/04/2023 and capabilities that include chat API and code.

Codey for Code Chat

codechat-bison is the name of the model that supports code chat. It's a foundation model that supports multi-turn conversations that are specialized for code. The model allows developers to chat with a chatbot for help with code-related questions. The code chat Codey API is used to interface with the codechat-bison model. The Codey APIs are in the PaLM API family.
The Codey API for code chat is ideal for code tasks that are completed with back-and-forth interactions so you can engage in a continuous conversation. For code tasks that require a single interaction, use the Codey API for code completion or the Codey API for code generation.

This document describes a software program called codechat-bison. It is a closed source text-based program that includes a chat API and code capabilities. The program has a maximum context of 6 and is not deprecated. It will be released on June 29, 2023. An access link to the program is available at https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/codechat-bison.

Embeddings for text is a closed source text embedding model that offers fast and accurate representation of text data. With a release date of 07/06/2023, this model includes pricing of $0.0001 per 1K characters and is not deprecated. Access link can be found at https://console.cloud.google.com/vertex-ai/publishers/google/model-garden/textembedding-gecko.

Embeddings for text

This document contains information about a text embedding model called "textembedding-gecko". It includes details such as the model's license, type, release date, and specifications such as its max context and vector size.

text-bison is a chat API LLM that offers closed source text capabilities with a maximum context of 8. It is published by Google Model Garden and will be released on 07/06/2023. The model is designed for use cases such as chatbots and conversation engines.

PaLM 2 for Text

text-bison is the name of the PaLM 2 for text large language model that understands and generates language. It's a foundation model that performs well at a variety of natural language tasks such as sentiment analysis, entity extraction, and content creation. The type of content that text-bison can create includes document summaries, answers to questions, and labels that classify content.
The PaLM 2 for text is ideal for tasks that can be completed with one API response, without the need for continuous conversation. For text tasks that require back-and-forth interactions, use the PaLM 2 for chat.

Text-bison is a chat API model with a maximum context of 8, released on 07/06/2023 and available on the Google Cloud Console. It is closed source and has capabilities in natural language processing.

Chat-bison is a powerful closed source language model designed for natural, multi-turn conversations. It excels in language understanding, generation and conversations, making it ideal for text tasks that require back-and-forth interactions. With capabilities for chat API, it has a maximum context of 4 and is perfect for a wide range of text-based applications.

PaLM 2 for Chat

The chat-bison model is a large language model that excels at language understanding, language generation and conversations. This chat model is fine-tuned to conduct natural multi-turn conversations.
The PaLM 2 for chat is ideal for text tasks that require back-and-forth interactions. For text tasks that can be completed with one API response (without the need for continuous conversation), use the PaLM 2 for text.

This document provides a brief description of the chat-bison software, including its license, release date, capabilities, and maximum context.

Flan T5 XXL is an open-source, high-performance text model with 11B parameters, achieving state-of-the-art performance on many benchmarks and fine-tuned on over 1000 additional tasks covering more languages. This model is licensed under Open Source and was released on 21/10/2022. With its impressive capabilities and strong few-shot performance, Flan T5 XXL is a powerful tool for natural language processing tasks of all kinds.

Flan T5 XXL

If you already know T5, FLAN-T5 is just better at everything. For the same number of parameters, these models have been fine-tuned on more than 1000 additional tasks covering also more languages. As mentioned in the first few lines of the abstract : Flan-PaLM 540B achieves state-of-the-art performance on several benchmarks, such as 75.2% on five-shot MMLU. We also publicly release Flan-T5 checkpoints,1 which achieve strong few-shot performance even compared to much larger models, such as PaLM 62B. Overall, instruction finetuning is a general method for improving the performance and usability of pretrained language models.

Flan T5 XXL is an open-source, high-performance text model with 11B parameters, achieving state-of-the-art performance on many benchmarks, and fine-tuned on over 1000 additional tasks covering more languages.

LaBSE is an open source, multilingual embedding model that supports 109 languages. With its Language-agnostic BERT architecture, it is capable of producing high-quality sentence embeddings for a wide range of applications.

LaBSE

Language-agnostic BERT sentence embedding model supporting 109 languages.

LaBSE is an open source, multilingual embedding model capable of producing high-quality sentence embeddings for a wide range of applications.

Google AI

This is a deprecated Closed Source text document with the ID text-curie-001. It has a max context of 2 and was released on 01/06/2020. No further information is available.

Text-curie-001 is the first version of the Curie model and has a capacity of 6.7 billion parameters. This model is designed to be more efficient than the larger models and can perform a variety of natural language processing tasks.

This is a deprecated closed-source text model released on 01/06/2020 with a maximum context of 2. No further information is available.

This document is a deprecated closed-source text file with the name "text-babbage-001", released on 01/06/2020.

This is a GPT-3 language model released in June 2020, named after the famous mathematician and inventor Charles Babbage. Text-babbage-001 has a capacity of 1.2 billion parameters and can perform a range of natural language processing tasks.

This is a deprecated closed-source text model with no description available.

This is a deprecated closed source text LLM with a release date of 01/07/2021. Its name is "text-ada-001" and it has a max context of 2. A short description of this LLM is not available as there is no information provided in the document about the actual content or capabilities of the LLM.

Capable of very simple tasks, usually the fastest model in the GPT-3 series, and lowest cost.

This is a deprecated closed-source text model with a maximum context of 2, released on 01/07/2021.

"gpt-3.5-turbo-16k" is an advanced language model that uses 16k parameters for improved performance. This document provides information about the model and a brief statement about the character named Sally in the text.

Same capabilities as the standard gpt-3.5-turbo model but with 4 times the context.

This document describes a closed-source text model called gpt-3.5-turbo-16k, which has capabilities including chat API, code, and multilingual support. Its maximum context is 16. The model was released on 13/06/2023 and is not currently deprecated.

Codex is a closed source text-based language model designed for evaluating large codebases. With its advanced capabilities and extensive training data, Codex provides a powerful tool for developers and researchers alike.

Codex

Evaluating Large Language Models Trained on Code

Codex is a closed source text-based language model designed for evaluating large codebases.

This is a powerful Codex model, capable of translating natural language to code and completing code. It supports completions within code as well. The model is closed source and has code capabilities.

Most capable Codex model. Particularly good at translating natural language to code. In addition to completing code, also supports inserting completions within code.

This document describes a Codex model with powerful natural language to code translation capabilities. It is closed source and deprecated.

text-embedding-ada-002 is a powerful embedding model for text search, text similarity, and code search. It outperforms OpenAI's previous most capable model, Davinci, while being priced 99.8% lower. With a vector size of 1536, it can handle up to 8 context windows and is available for $0.0004 per 1k tokens. This closed-source embedding model was released on December 15, 2022, and is ideal for large-scale natural language processing tasks.

This model replaces five separate models for text search, text similarity, and code search, and outperforms our previous most capable model, Davinci, at most tasks, while being priced 99.8% lower.

text-embedding-ada-002 is a powerful closed-source embedding model for text search, text similarity, and code search that outperforms OpenAI's previous most capable model, Davinci, while being priced 99.8% lower. It has a vector size of 1536 and can handle up to 8 context windows.

text-davinci-003 is a closed source text-based language model released by OpenAI on 01/06/2020. Capable of code, instruction, and multilingual processing, this LLM has a max context of 4. However, it has been deprecated.

GPT 3

Can do any language task with better quality, longer output, and consistent instruction-following than the curie, babbage, or ada models. Also supports some additional features such as inserting text.

text-davinci-003 is a closed source language model developed by OpenAI with the ability to code, give instructions, and work with multiple languages, and has a maximum context of 4. It was announced on January 6, 2020, with the release of GPT-4, but is now deprecated.

text-davinci-002 is a deprecated closed source language model developed by OpenAI. It was released on 01/07/2021 and has a maximum context of 4. Its capabilities are not specified.

Similar capabilities totext-davinci-003but trained with supervised fine-tuning instead of reinforcement learning

text-davinci-002 is a closed source text generation model with a max context of 4, released on 01/07/2021 and is now deprecated.

GPT 4 32k is a closed-source AI text model with 4x the context length of the base GPT-4 model. It has the same capabilities as the base model and is updated with the latest iteration. The model was trained up to September 2021 and is available through a chat API. Pricing is based on prompt and sampled tokens and is $0.06/1k and $0.12/1k respectively. The model was released with a license and has a maximum context of 32.

GPT 4 32k

Same capabilities as the base gpt-4 mode but with 4x the context length. Will be updated with our latest model iteration.

This document describes the GPT-4-32k model, a closed-source text-based AI with various capabilities, including chat API, code, instruct, multilingual, and multimodal. It has a maximum context of 32 and is not deprecated.

GPT 3.5 Turbo is an affordable and powerful chat-optimized AI model with multilingual capabilities. It is the most capable GPT-3.5 model and optimized for chat, at 1/10th the cost of text-davinci-003. With training data up to September 2021 and a release date of January 3, 2023, this closed source model offers input pricing at $0.0015 per 1K tokens and output pricing at $0.002 per 1K tokens, making it an ideal choice for chat API and multilingual capabilities with a max context of 4.

GPT 3.5 Turbo

Most capable GPT-3.5 model and optimized for chat at 1/10th the cost of text-davinci-003. Will be updated with our latest model iteration. We recommend using gpt-3.5-turbo over the other GPT-3.5 models because of its lower cost.

GPT 3.5 Turbo is an affordable and powerful chat-optimized AI model with multilingual capabilities.

GPT 4 is OpenAI's latest advanced text generation model, optimized for chat and capable of complex tasks. It is multilingual and multimodal, with a max context of 8. It was released on March 14th, 2023, and is licensed as closed source. GPT 4 uses training data up to September 2021 and is priced at $0.03 per 1k prompt tokens and $0.06 per 1k sampled tokens.

GPT 4

More capable than any GPT-3.5 model, able to do more complex tasks, and optimized for chat. Will be updated with our latest model iteration.

GPT-4 is OpenAI's latest advanced text generation model optimized for chat and capable of complex tasks.

OpenAI

Writer is an AI writing platform for teams.

Palmyra Large is a closed source text AI model with a size of 20B. It provides advanced natural language processing capabilities and is suitable for various text analysis tasks. Its release date is not specified, and pricing data is not available.

Palmyra Large

Palmyra Large is a closed source text AI model with advanced natural language processing capabilities.

Camel-5b is a 5B Open Source large language model trained on 70K instruction-response records. Its state-of-the-art architecture delivers exceptional performance and versatility for advanced natural language processing and comprehension capabilities. With its meticulous training and fine-tuning techniques, Camel-5b offers unparalleled proficiency in executing language-based instructions.

Camel

Camel-5b, a state-of-the-art instruction-following large language model designed to deliver exceptional performance and versatility. Derived from the foundational architecture of Palmyra-Base, Camel-5b is specifically tailored to address the growing demand for advanced natural language processing and comprehension capabilities.

This is a document describing a software project called "camel-5b-hf". It is an open source text project with an access link to huggingface.co. The release date is set for January 4th, 2023, and there is a demo link available at chatcamel.vercel.app. The project is not deprecated.

PalmyraMed is a closed-source text-based AI model, fine-tuned on a custom-curated medical dataset of 200,000 examples. Palmyra-Med-20b and Palmyra-Med-40b outperform other LLMs in the medical domain, as shown by their performance on multiple medical knowledge datasets.

PalmyraMed

By applying instruction-based fine-tuning on a custom-curated medical dataset of 200,000 examples, we create novel, fine-tuned models, Palmyra-Med-20b and Palmyra-Med-40b. Performance is then measured across multiple medical knowledge datasets, including PubMedQA and MedQA. Our fine-tuned models outperform both their base counterparts and other LLMs pretrained on domain-specific knowledge. This research demonstrates the effectiveness of instruction-based fine-tuning in enhancing LLMs performance in the medical domain.

PalmyraMed is a closed-source text-based AI model for the medical domain, fine-tuned on a custom-curated dataset of 200,000 examples.

Palmyra E is a closed source text model boasting 30B parameters, available on-premise or via an API. Its three flavors, General, Healthcare, and Fintech, ensure it is more capable than GPT-3 and GPT-3.5 models, and it is capable of performing complex tasks.

Palmyra E

It is more capable than the GPT-3 and GPT-3.5 models, able to perform more complex tasks, and comes in three flavors: General, Healthcare, and Fintech. It is available on-premise or via an API.

Palmyra E is a closed source text model with 30B parameters available on-premise or via an API, capable of performing complex tasks across various fields.

Silk Road is a closed source text-based AI model with a maximum context length of 85 characters. It offers the same capabilities as the Palmyra-E mode, but with an increased context length of ~80K. This model is suitable for a wide range of natural language processing tasks and is available for use under a closed source license.

Silk Road

Same capabilities as the Palmyra-E mode but with ~80K context length.

Silk Road is a closed-source text-based AI model suitable for natural language processing tasks with a maximum context length of 85 characters and an increased context length of ~80K.

Palmyra X is a text-based model capable of comprehending the intent of text, solving logic problems, and summarizing audiences. With a size of 43B and impressive capabilities in few-shots, cause and effect, and audience summarization, Palmyra X is a powerful tool for instruction. Closed source and non-deprecated, this model is ready for use.

Palmyra X

Palmyra Large is the most capable model family, capable of performing any task that the other models can, often with less instruction. Palmyra Large is good at comprehending the text's intent, solving logic problems, and explaining character motivations.
Good at: Few-shots, cause and effect, and audience summarization

Palmyra X is a powerful closed-source text-based model capable of comprehending intent, solving logic problems, and summarizing audiences.

Palmyra Base is an open source, text-based language model primarily pre-trained with English text and utilizing a causal language modeling (CLM) objective. With a release date of 13/02/2023, Palmyra Base provides a robust set of capabilities for natural language processing tasks. Its training corpus includes CommonCrawl data, and it is available for free through the Hugging Face access link.

Palmyra Base

Palmyra Base was primarily pre-trained with English text. Note that there is still a trace amount of non-English data present within the training corpus that was accessed through CommonCrawl. A causal language modeling (CLM) objective was utilized during the process of the model's pretraining. Similar to GPT-3, Palmyra Base is a member of the same family of models that only contain a decoder. As a result, it was pre-trained utilizing the objective of self-supervised causal language modeling. Palmyra Base uses the prompts and general experimental setup from GPT-3 in order to conduct its evaluation per GPT-3.

Palmyra Base is an open source, text-based language model primarily pre-trained with English text and utilizing a causal language modeling (CLM) objective.

Writer

Luminous Control Models are fine-tuned for a diverse set of tasks described by human-written instructions, resulting in better zero-shot performance and lower costs. Capable of instructing and multilingual, these models come in base, extended, and supreme sizes and are licensed as closed source.

Luminous Control Models

Our control models are versions of models optimized to follow instructions. This means that these models have been fine-tuned for a diverse set of tasks described by human-written instructions. One of their great advantages is that they have much better zero-shot performance compared to our vanilla models. That makes them easier to use and also leads to fewer tokens in the input, resulting in lower costs. We provide a control version of a model for all three model sizes: base, extended, and supreme.

Luminous Control Models are optimized to follow human-written instructions and come in base, extended, and supreme sizes. They have better zero-shot performance and are multilingual.

Luminous Control Models are fine-tuned versions of text models designed to follow human-written instructions. With better zero-shot performance and lower input costs, these models come in base, extended, and supreme sizes and can instruct and handle multiple languages. Available at a cost of $0.0375-$0.05625/1k input token and $0.03024 per input image, they are closed source and not deprecated.

Luminous Control Models are optimized versions of text models designed to follow human-written instructions, with better zero-shot performance and lower input costs.

Luminous Control Models: optimized for human-written instructions, with better zero-shot performance and lower costs. Multilingual and capable of following instructions. Closed source.

This document describes the "Luminous Control Models," which are AI models optimized for following human-written instructions. They offer better zero-shot performance and lower costs, and are available in all three model sizes. Additionally, they are capable of instructing and completing multilingual tasks.

Luminous Explore is a multilingual model for world-class semantic representation, based on Luminous-base. It offers high-quality embeddings for a variety of applications, with closed-source licensing. Check out our announcement link for more information.

Luminous Explore

Luminous-Explore is our model for semantic embeddings based on Luminous-base.

Luminous Explore is a multilingual semantic embedding model with high-quality embeddings for various applications.

Luminous Supreme is a powerful text generation model with multilingual capabilities. This closed source model is 70B in size and can solve all tasks of smaller models, making it perfect for creative writing. Pricing is at $0.175/1k input token.

Luminous Supreme

Luminous-supreme is the largest model in the Luminous model family and consequently the most capable one. It can solve all tasks the smaller models can solve as well and is especially well suited for creative text writing.

Luminous Supreme is a powerful 70B text generation model with multilingual capabilities.

Luminous Extended is a multilingual, multimodal 30B text model for information extraction and language simplification, available under a closed source license. It performs better on a wide range of tasks compared to Luminous-base and is faster and cheaper than Luminous-supreme.

Luminous Extended

Luminous-extended is our second-largest model. It is well suited for tasks like information extraction and language simplification. It performs better on a wide range of tasks compared to Luminous-base and is faster and cheaper than Luminous-supreme.

This document provides information about Luminous Extended, a 30B text model that is multilingual and multimodal, and performs well on tasks such as information extraction and language simplification. It is faster and cheaper than Luminous Supreme, and is available under a closed source license.

Luminous Base is a closed-source text-based model with multilingual and multimodal capabilities, designed for classification and labeling tasks. It is fast, affordable, and has a size of 13B. Pricing is $0.03/1k input token and $0.03024/input image. Release date and training data are not specified.

Luminous Base

Luminous-base is our smallest model. That makes it the fastest and cheapest model in the Luminous model family. Therefore, this model is suited for applications where speed is important and costs should be low. It is very well suited for tasks like classification and labelling.

# luminous-base

Luminous Base is a text-based AI model that is ideal for classification and labeling tasks where speed and cost-efficiency are key. It is the smallest and most affordable model in the Luminous family, with multilingual and multimodal capabilities, and a compact size of just 13B. Pricing starts at $0.03/1k input token and $0.03024/per input image. Luminous Base is closed source and not deprecated.

Aleph Alpha

llama7b-v2-chat is a multilingual LLM with limited use license for text data. It has a release date of 18/07/2023 and is not deprecated. This LLM is suitable for chat applications and has a compact size.

LLama 2 (7B) Chat

llama7b-v2-chat is a multilingual text-based model with limited use license, set to release on 18/07/2023.

llama13b-v2-chat is a multilingual LLM with limited use license and text type. It is set to be released on 18/07/2023 and is not deprecated. This LLM provides a chat capability and is suitable for various use cases.

LLama 2 (13B) Chat

This is a multilingual chat model with limited use license.

LLama 2 (70B) Chat is an open-source, multilingual text model with 70B parameters, significant improvements over Llama 1 models, and fast inference capabilities.

LLama 2 (70B) Chat

The Llama 2 release introduces a family of pretrained and fine-tuned LLMs, ranging in scale from 7B to 70B parameters (7B, 13B, 70B). The pretrained models come with significant improvements over the Llama 1 models, including being trained on 40% more tokens, having a much longer context length (4k tokens 🤯), and using grouped-query attention for fast inference of the 70B model🔥!

This document appears to be related to a chat tool called llama70b-v2. It includes information such as the license, release date, and capabilities of the tool, as well as an announcement link. The document also notes that the tool is not deprecated and has a maximum context of 4, but does not provide any additional information beyond that.

Galactica is a text-based AI model trained on a large-scale scientific corpus, capable of performing a wide range of scientific tasks, including citation prediction, QA, mathematical reasoning, summarization, document generation, molecular property prediction, and entity extraction.

Galactica

The GALACTICA models are trained on a large-scale scientific corpus. The models are designed to perform scientific tasks, including but not limited to citation prediction, scientific QA, mathematical reasoning, summarization, document generation, molecular property prediction and entity extraction.

Galactica is a text-based AI model trained on a large-scale scientific corpus capable of performing various scientific tasks.

OPT (Open Pre-trained Transformers) is an open-source suite of decoder-only transformers with 175B parameters. It is designed to support responsible research and development by providing pre-trained models to interested researchers.

Open Pre-trained Transformers (175B parameters) is a suite of decoder-only pre-trained transformers that are fully and responsibly shared with interested researchers

OPT (Open Pre-trained Transformers) is an open-source suite of decoder-only transformers with 175B parameters, designed to support responsible research and development.

OPT-IML is an open source text type model with instruct capabilities. It is optimized using instruction-tuned versions of OPT and trained on OPT-IML Bench, a large benchmark for Instruction MetaLearning (IML) of 2000 NLP tasks. It has a size of 175B, released on 22/12/2022 and is licensed under the Open Source license.

OPT IML

OPT-IML (OPT + Instruction Meta-Learning) is a set of instruction-tuned versions of OPT, on a collection of ~2000 NLP tasks gathered from 8 NLP benchmarks, called OPT-IML Bench.

OPT-IML is an open source text type model that is optimized using instruction-tuned versions of OPT and trained on OPT-IML Bench, a large benchmark for Instruction MetaLearning (IML) of 2000 NLP tasks. It has a size of 175B, released on 22/12/2022 and is licensed under the Open Source license.

Llama is an open source text model capable of generating short descriptions of less than 300 characters based on various inputs including training data, size, license, release date, and capabilities.

Llama

This document provides information about the Llama-13b open source text model, including its capabilities, size, and license, as well as some example questions and answers.

Small

Small is a closed-source text-based AI assistant that is suitable for personal or professional use.

Command-medium is a closed source text editor that enables users to edit text and code. It has capabilities for editing text and code, and is licensed as closed source. The software has no specific pricing data and was released with no specified release date.

Medium

This document provides information about the text editor Medium, including its features, licensing, and pricing, as well as a set of math and logic puzzles.

Large is a closed source text-based command designed for efficient data processing. Its capabilities include analyzing large datasets, automating tasks, and integrating with other software. Large is designed to optimize performance and streamline workflows for businesses and organizations.

Large

Large is a text-based command for efficient data processing, designed to optimize performance and streamline workflows for businesses and organizations.

Extremely Large is a closed source text-based tool with a size of 52.4B. It offers powerful capabilities for a wide range of use cases. Learn more about this tool and its beta launch in the announcement link.

Extremely Large

Command-xlarge is a closed-source, extremely large text AI model developed by Cohere, with advanced capabilities for text processing and analysis.

embed-english-light-v2.0 is a closed-source embedding model with a vector size of 768. Its smaller "light" size allows for faster performance, while larger models provide better accuracy.

Smaller "light" models are faster, while larger models will perform better.

embed-english-light-v2.0 is a closed-source embedding model with a vector size of 768, designed for faster performance.

embed-english-v2.0 is a closed source embedding model with a vector size of 4096. It provides fast and efficient language processing capabilities.

embed-english-v2.0 is a closed source language embedding model with a vector size of 4096.

embed-multilingual-v2.0 is a closed-source embedding model capable of handling multiple languages using a vector size of 768. With no deprecation, the model can be used for various embedding tasks.

embed-multilingual-v2.0 is a closed-source embedding model capable of handling multiple languages using a vector size of 768.

Company	Founded	Website	Models
Google AI	🇺🇸 2017	Website	code-gecko PaLM-E Flan-T5/PaLM UL2 Palm Minerva Lamda Flan T5 Codey for Code Generation Codey for Code Chat Embeddings for text PaLM 2 for Text PaLM 2 for Chat Flan T5 XXL LaBSE
OpenAI	🇺🇸 2015	Website	text-curie-001 text-babbage-001 text-ada-001 gpt-3.5-turbo-16k Codex code-davinci-002 text-embedding-ada-002 GPT 3 text-davinci-002 GPT 4 32k GPT 3.5 Turbo GPT 4
Writer	🇺🇸	Website	Palmyra Large Camel PalmyraMed Palmyra E Silk Road Palmyra X Palmyra Base
Aleph Alpha	🇩🇪		Luminous Control Models Luminous Control Models Luminous Control Models Luminous Explore Luminous Supreme Luminous Extended Luminous Base
Meta	🇺🇸	Website	LLama 2 (7B) Chat LLama 2 (13B) Chat LLama 2 (70B) Chat Galactica OPT OPT IML Llama
Cohere	🇨🇦 2019	Website	Small Medium Large Extremely Large embed-english-light-v2.0 embed-english-v2.0 embed-multilingual-v2.0
Tsinghua University		Website	CodeGeeX2 ChatGLM2-6B ChatGLM-3B ChatGLM CodeGeeX GLM-130b
EleutherAI	🇺🇸 2020	Website	Pythia GPT NeoXT Chat Base (20B)GPT NeoX GPT J GPT Neo
AI21 Labs	🇮🇱 2017	Website	Jurassic-2 Jumbo-Instruct Jurassic-2 Grande-Instruct Jurassic-2 Large Jurassic-2 Grande Jurassic-2 Jumbo
Anthrophic	🇺🇸 2021	Website	Claude V1 (100k)Claude Instant V1 (100k)Claude 2 Claude Instant V1 Claude V1
StabilityAI			StableCode Instruct Alpha 3B StableCode Completion Alpha 3B Stable Beluga 2 StableLM-Tuned-Alpha
Salesforce AI	🇺🇸	Website	CodeT5+ 16B XGen-7B CodeGen
MosaicML	🇺🇸	Website	MPT-30B-Instruct MPT-30B MPT-30B-Chat
BigCode	🇺🇸	Website	StarCoder Plus StarCoder SantaCoder
			WizardCoder 15B 1.0 WizardLM-13B V1.2
UAE TII		Website	Falcon 40B Instruct Falcon
BigScience	🇺🇸 2021	Website	Bloomz Bloom
			oasst-sft-4-pythia-12b-epoch-3.5
DataBricks	🇺🇸	Website	Dolly V2
CalderaAI	2023	Website	30B-Lazarus
Yandex		Website	YaLM
Microsoft AI	🇺🇸	Website	Kosmos-1
Baidu			ERNIE Bot
NVIDIA	🇺🇸		Megatron-Turing NLG
Stanford	🇺🇸	Website	Alpaca 7B
Huawei

Organisations building LLM models