# Model Collection

import { Callout, FileTree } from 'nextra-theme-docs'

<Callout emoji="⚠️">
  This section is under heavy development.
</Callout>

This section consists of a collection and summary of notable and foundational LLMs. (Data adopted from [Papers with Code](https://paperswithcode.com/methods/category/language-models) and the recent work by [Zhao et al. (2023)](https://arxiv.org/pdf/2303.18223.pdf).

## Models

| Model | Release Date | Description | 
| --- | --- | --- | 
| [BERT](https://arxiv.org/abs/1810.04805)| 2018 | Bidirectional Encoder Representations from Transformers | 
| [GPT](https://s3-us-west-2.amazonaws.com/openai-assets/research-covers/language-unsupervised/language_understanding_paper.pdf) | 2018 | Improving Language Understanding by Generative Pre-Training | 
| [RoBERTa](https://arxiv.org/abs/1907.11692) | 2019 | A Robustly Optimized BERT Pretraining Approach | 
| [GPT-2](https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf) | 2019 | Language Models are Unsupervised Multitask Learners | 
| [T5](https://arxiv.org/abs/1910.10683) | 2019 | Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer | 
| [BART](https://arxiv.org/abs/1910.13461) | 2019 | Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension |
| [ALBERT](https://arxiv.org/abs/1909.11942) |2019 | A Lite BERT for Self-supervised Learning of Language Representations | 
| [XLNet](https://arxiv.org/abs/1906.08237) | 2019 | Generalized Autoregressive Pretraining for Language Understanding and Generation |
| [CTRL](https://arxiv.org/abs/1909.05858) |2019 | CTRL: A Conditional Transformer Language Model for Controllable Generation | 
| [ERNIE](https://arxiv.org/abs/1904.09223v1) | 2019| ERNIE: Enhanced Representation through Knowledge Integration |
| [GShard](https://arxiv.org/abs/2006.16668v1) | 2020 | GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding |
| [GPT-3](https://arxiv.org/abs/2005.14165) | 2020 | Language Models are Few-Shot Learners |
| [LaMDA](https://arxiv.org/abs/2201.08239v3) | 2021 | LaMDA: Language Models for Dialog Applications | 
| [PanGu-α](https://arxiv.org/abs/2104.12369v1) | 2021 | PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation |
| [mT5](https://arxiv.org/abs/2010.11934v3) | 2021 | mT5: A massively multilingual pre-trained text-to-text transformer |
| [CPM-2](https://arxiv.org/abs/2106.10715v3) | 2021 | CPM-2: Large-scale Cost-effective Pre-trained Language Models |
| [T0](https://arxiv.org/abs/2110.08207) |2021 |Multitask Prompted Training Enables Zero-Shot Task Generalization |
| [HyperCLOVA](https://arxiv.org/abs/2109.04650) | 2021 | What Changes Can Large-scale Language Models Bring? Intensive Study on HyperCLOVA: Billions-scale Korean Generative Pretrained Transformers |
| [Codex](https://arxiv.org/abs/2107.03374v2) |2021 |Evaluating Large Language Models Trained on Code |
| [ERNIE 3.0](https://arxiv.org/abs/2107.02137v1) | 2021 | ERNIE 3.0: Large-scale Knowledge Enhanced Pre-training for Language Understanding and Generation|
| [Jurassic-1](https://uploads-ssl.webflow.com/60fd4503684b466578c0d307/61138924626a6981ee09caf6_jurassic_tech_paper.pdf) | 2021 | Jurassic-1: Technical Details and Evaluation |
| [FLAN](https://arxiv.org/abs/2109.01652v5) | 2021 | Finetuned Language Models Are Zero-Shot Learners |
| [MT-NLG](https://arxiv.org/abs/2201.11990v3) | 2021 | Using DeepSpeed and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model|
| [Yuan 1.0](https://arxiv.org/abs/2110.04725v2) | 2021| Yuan 1.0: Large-Scale Pre-trained Language Model in Zero-Shot and Few-Shot Learning |
| [WebGPT](https://arxiv.org/abs/2112.09332v3) | 2021 | WebGPT: Browser-assisted question-answering with human feedback |
| [Gopher](https://arxiv.org/abs/2112.11446v2) |2021 | Scaling Language Models: Methods, Analysis & Insights from Training Gopher |
| [ERNIE 3.0 Titan](https://arxiv.org/abs/2112.12731v1) |2021 | ERNIE 3.0 Titan: Exploring Larger-scale Knowledge Enhanced Pre-training for Language Understanding and Generation |
| [GLaM](https://arxiv.org/abs/2112.06905) | 2021 | GLaM: Efficient Scaling of Language Models with Mixture-of-Experts |
| [InstructGPT](https://arxiv.org/abs/2203.02155v1) | 2022 | Training language models to follow instructions with human feedback |
| [GPT-NeoX-20B](https://arxiv.org/abs/2204.06745v1) | 2022 | GPT-NeoX-20B: An Open-Source Autoregressive Language Model |
| [AlphaCode](https://arxiv.org/abs/2203.07814v1) | 2022 | Competition-Level Code Generation with AlphaCode |
| [CodeGen](https://arxiv.org/abs/2203.13474v5) | 2022 | CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis |
| [Chinchilla](https://arxiv.org/abs/2203.15556) | 2022 | Shows that for a compute budget, the best performances are not achieved by the largest models but by smaller models trained on more data. |
| [Tk-Instruct](https://arxiv.org/abs/2204.07705v3) | 2022 | Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks |
| [UL2](https://arxiv.org/abs/2205.05131v3) | 2022 | UL2: Unifying Language Learning Paradigms |
| [PaLM](https://arxiv.org/abs/2204.02311v5) |2022| PaLM: Scaling Language Modeling with Pathways |
| [OPT](https://arxiv.org/abs/2205.01068) | 2022 | OPT: Open Pre-trained Transformer Language Models |
| [BLOOM](https://arxiv.org/abs/2211.05100v3) | 2022 | BLOOM: A 176B-Parameter Open-Access Multilingual Language Model |
| [GLM-130B](https://arxiv.org/abs/2210.02414v1) | 2022 | GLM-130B: An Open Bilingual Pre-trained Model |
| [AlexaTM](https://arxiv.org/abs/2208.01448v2) | 2022 | AlexaTM 20B: Few-Shot Learning Using a Large-Scale Multilingual Seq2Seq Model |
| [Flan-T5](https://arxiv.org/abs/2210.11416v5) | 2022 | Scaling Instruction-Finetuned Language Models |
| [Sparrow](https://arxiv.org/abs/2209.14375) | 2022 | Improving alignment of dialogue agents via targeted human judgements |
| [U-PaLM](https://arxiv.org/abs/2210.11399v2) | 2022 | Transcending Scaling Laws with 0.1% Extra Compute |
| [mT0](https://arxiv.org/abs/2211.01786v1) | 2022 | Crosslingual Generalization through Multitask Finetuning |
| [Galactica](https://arxiv.org/abs/2211.09085v1) | 2022 | Galactica: A Large Language Model for Science |
| [OPT-IML](https://arxiv.org/abs/2212.12017v3) | 2022 | OPT-IML: Scaling Language Model Instruction Meta Learning through the Lens of Generalization |
| [LLaMA](https://arxiv.org/abs/2302.13971v1) | 2023 | LLaMA: Open and Efficient Foundation Language Models |
| [GPT-4](https://arxiv.org/abs/2303.08774v3) | 2023 |GPT-4 Technical Report |
| [PanGu-Σ](https://arxiv.org/abs/2303.10845v1) | 2023 | PanGu-Σ: Towards Trillion Parameter Language Model with Sparse Heterogeneous Computing |
| [BloombergGPT](https://arxiv.org/abs/2303.17564v1)| 2023 |BloombergGPT: A Large Language Model for Finance|
| [PaLM 2](https://ai.google/static/documents/palm2techreport.pdf) | 2023 | A Language Model that has better multilingual and reasoning capabilities and is more compute-efficient than its predecessor PaLM. |