[Docs] Iterate model prebuilts docs (mlc-ai#1043)

CharlieFRuan · web-flow · commit a032d40bfc4b · 2023-10-09T15:03:24.000-07:00
* Iterate model prebuilts docs

* small fix
diff --git a/README.md b/README.md
@@ -52,8 +52,26 @@ Machine Learning Compilation for Large Language Models (MLC LLM) is a high-perfo
   </tbody>
 </table>
 
-**Prebuilt model support.** MLC LLM supports a wide range of model architectures and variants. We have the following prebuilts which you can
-use off-the-shelf. Visit [Prebuilt Models](https://llm.mlc.ai/docs/prebuilt_models.html) to see the full list.
+## News
+
+* [08/25/2023] CodeLlama support is up.
+* [08/14/2023] [[Post]](https://blog.mlc.ai/2023/08/09/GPU-Accelerated-LLM-on-Orange-Pi) Mali GPU support is up on Orange Pi.
+* [08/09/2023] [[Post]](https://blog.mlc.ai/2023/08/09/Making-AMD-GPUs-competitive-for-LLM-inference) ROCm backend is mature to use.
+* [08/02/2023] [Dockerfile](https://github.com/mlc-ai/llm-perf-bench/) is released for CUDA performance benchmarking.
+* [07/19/2023] Support for Llama2-7B/13B/70B is up.
+* [05/22/2023] [[Post]](https://blog.mlc.ai/2023/05/22/bringing-open-large-language-models-to-consumer-devices) RedPajama support is up.
+* [05/08/2023] [[Post]](https://blog.mlc.ai/2023/05/08/bringing-hardware-accelerated-language-models-to-android-devices) MLC LLM is now available on Android.
+* [05/01/2023] [[Post]](https://blog.mlc.ai/2023/05/01/bringing-accelerated-llm-to-consumer-hardware) MLC LLM is released with Metal, Vulkan and CUDA backends.
+* [04/14/2023] [WebLLM](https://github.com/mlc-ai/web-llm) is released prior to MLC LLM with WebGPU and WebAssembly backend.
+
+## Getting Started
+
+Please visit our [documentation](https://llm.mlc.ai/docs/index.html#getting-started) for detailed instructions.
+
+## Prebuilt model support
+
+MLC LLM supports a wide range of model architectures and variants. We have the following prebuilts which you can
+use off-the-shelf. Visit [Prebuilt Models](https://llm.mlc.ai/docs/prebuilt_models.html) to see the full list, and [Compile Models via MLC](https://llm.mlc.ai/docs/compilation/compile_models.html) to see how to use models not on this list.
 
 <table style="width:100%">
   <thead>
@@ -64,29 +82,8 @@ use off-the-shelf. Visit [Prebuilt Models](https://llm.mlc.ai/docs/prebuilt_mode
   </thead>
   <tbody>
     <tr>
-      <td rowspan=8>Llama</td>
-      <td>Llama-2</td>
-    </tr>
-    <tr>
-      <td>Code Llama</td>
-    </tr>
-    <tr>
-      <td>Vicuna</td>
-    </tr>
-    <tr>
-      <td>WizardLM</td>
-    </tr>
-    <tr>
-      <td>WizardMath</td>
-    </tr>
-    <tr>
-      <td>OpenOrca Platypus2</td>
-    </tr>
-    <tr>
-      <td>FlagAlpha Llama-2 Chinese</td>
-    </tr>
-    <tr>
-      <td>georgesung Llama-2 Uncensored</td>
+      <td>Llama</td>
+      <td>Llama-2, Code Llama, Vicuna, WizardLM, WizardMath, OpenOrca Platypus2, FlagAlpha Llama-2 Chinese, georgesung Llama-2 Uncensored</td>
     </tr>
     <tr>
       <td>GPT-NeoX</td>
@@ -112,25 +109,13 @@ use off-the-shelf. Visit [Prebuilt Models](https://llm.mlc.ai/docs/prebuilt_mode
       <td>ChatGLM</td>
       <td></td>
     </tr>
+    <tr>
+      <td>StableLM</td>
+      <td></td>
+    </tr>
   </tbody>
 </table>
 
-## News
-
-* [08/25/2023] CodeLlama support is up.
-* [08/14/2023] [[Post]](https://blog.mlc.ai/2023/08/09/GPU-Accelerated-LLM-on-Orange-Pi) Mali GPU support is up on Orange Pi.
-* [08/09/2023] [[Post]](https://blog.mlc.ai/2023/08/09/Making-AMD-GPUs-competitive-for-LLM-inference) ROCm backend is mature to use.
-* [08/02/2023] [Dockerfile](https://github.com/mlc-ai/llm-perf-bench/) is released for CUDA performance benchmarking.
-* [07/19/2023] Support for Llama2-7B/13B/70B is up.
-* [05/22/2023] [[Post]](https://blog.mlc.ai/2023/05/22/bringing-open-large-language-models-to-consumer-devices) RedPajama support is up.
-* [05/08/2023] [[Post]](https://blog.mlc.ai/2023/05/08/bringing-hardware-accelerated-language-models-to-android-devices) MLC LLM is now available on Android.
-* [05/01/2023] [[Post]](https://blog.mlc.ai/2023/05/01/bringing-accelerated-llm-to-consumer-hardware) MLC LLM is released with Metal, Vulkan and CUDA backends.
-* [04/14/2023] [WebLLM](https://github.com/mlc-ai/web-llm) is released prior to MLC LLM with WebGPU and WebAssembly backend.
-
-## Getting Started
-
-Please visit our [this page](https://llm.mlc.ai/docs/index.html#getting-started) for detailed instructions.
-
 ## Universal Deployment APIs
 
 MLC LLM provides multiple sets of APIs across platforms and environments. These include
diff --git a/docs/compilation/distribute_compiled_models.rst b/docs/compilation/distribute_compiled_models.rst
@@ -161,7 +161,7 @@ Download the Distributed Models and Run in iOS App
 
 For iOS app, model libraries are statically packed into the app at the time of app building.
 Therefore, the iOS app supports running any models whose model libraries are integrated into the app.
-You can check the :ref:`list of supported model libraries <prebuilt-models-ios>`.
+You can check the :ref:`list of supported model libraries <using-prebuilt-models-ios>`.
 
 To download and run the compiled RedPajama-3B instruct model on iPhone, we need to reuse the integrated ``RedPajama-INCITE-Chat-3B-v1-q4f16_1`` model library.
 Please revisit :ref:`distribute-model-step3-specify-model-lib` and make sure the ``model_lib`` field of `mlc-chat-config.json` is set to ``RedPajama-INCITE-Chat-3B-v1-q4f16_1``.
diff --git a/docs/prebuilt_models.rst b/docs/prebuilt_models.rst
@@ -256,16 +256,13 @@ MLC-LLM supports the following model architectures:
   :widths: 10 10 15 15
   :header-rows: 1
 
-  * - Code
-    - Architecture
-    - Variants w/ MLC prebuilts
-    - Variants w/o MLC prebuilts
-  * - ``llama``
-    - LLaMa
-
-      * :ref:`Prebuilt library table <llama_library_table>`
-      * `Official link <https://github.com/facebookresearch/llama>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/llama.py>`__
+  * - Model Architecture
+    - Support
+    - Available MLC Prebuilts
+    - Unavailable in MLC Prebuilts
+  * - `LLaMA <https://github.com/facebookresearch/llama>`__
+    - * :ref:`Prebuilt Model Library <llama_library_table>`
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/llama.py>`__
     - * :ref:`Llama-2 <llama2_variant_table>`
       * :ref:`Code Llama <code_llama_variant_table>`
       * :ref:`Vicuna <vicuna_variant_table>`
@@ -280,64 +277,51 @@ MLC-LLM supports the following model architectures:
       * `Gorilla <https://huggingface.co/gorilla-llm/gorilla-7b-hf-delta-v0>`__
       * `YuLan-Chat <https://github.com/RUC-GSAI/YuLan-Chat>`__
       * `WizardCoder (new) <https://github.com/nlpxucan/WizardLM/tree/main/WizardCoder>`__
-  * - ``gpt-neox``
-    - GPT-NeoX 
-
-      * :ref:`Prebuilt library table <gpt_neox_library_table>`
-      * `Official link <https://github.com/EleutherAI/gpt-neox>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gpt_neox.py>`__
+  * - `GPT-NeoX <https://github.com/EleutherAI/gpt-neox>`__
+    - * :ref:`Prebuilt Model Library <gpt_neox_library_table>`
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gpt_neox.py>`__
     - * :ref:`RedPajama <red_pajama_variant_table>` 
     - * `Dolly <https://github.com/databrickslabs/dolly>`__
       * `Pythia <https://huggingface.co/EleutherAI/pythia-1.4b>`__
       * `StableCode <https://huggingface.co/stabilityai/stablecode-instruct-alpha-3b>`__
-  * - ``gptj``
-    - GPT-J
-
-      * Prebuilt not compiled yet
-      * `Official link <https://huggingface.co/EleutherAI/gpt-j-6b>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gptj.py>`__
+  * - `GPT-J <https://huggingface.co/EleutherAI/gpt-j-6b>`__
+    - * Prebuilt not compiled yet
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gptj.py>`__
     - 
     - * `MOSS <https://github.com/OpenLMLab/MOSS>`__
-  * - ``rwkv``
-    - RWKV 
-
-      * :ref:`Prebuilt library table <rwkv_library_table>`
-      * `Official link <https://github.com/BlinkDL/RWKV-LM>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/rwkv.py>`__
+  * - `RWKV <https://github.com/BlinkDL/RWKV-LM>`__
+    - * :ref:`Prebuilt Model Library <rwkv_library_table>`
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/rwkv.py>`__
     - * :ref:`RWKV-raven <rwkv_raven_variant_table>` 
     - 
-  * - ``minigpt``
-    - MiniGPT
-
-      * Prebuilt not compiled yet
-      * `Official link <https://huggingface.co/Vision-CAIR/MiniGPT-4>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/minigpt.py>`__
+  * - `MiniGPT <https://huggingface.co/Vision-CAIR/MiniGPT-4>`__
+    - * Prebuilt not compiled yet
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/minigpt.py>`__
     - 
     - * `MiniGPT-4 <https://huggingface.co/Vision-CAIR/MiniGPT-4>`__
-  * - ``gpt_bigcode``
-    - GPTBigCode
-
-      * :ref:`Prebuilt library table <gpt_big_code_library_table>`
-      * `Official link <https://huggingface.co/docs/transformers/model_doc/gpt_bigcode>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gpt_bigcode.py>`__
+  * - `GPTBigCode <https://huggingface.co/docs/transformers/model_doc/gpt_bigcode>`__
+    - * :ref:`Prebuilt Model Library <gpt_big_code_library_table>`
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/gpt_bigcode.py>`__
     - * :ref:`WizardCoder (old) <wizard_coder_variant_table>` 
     - * `StarCoder <https://huggingface.co/bigcode/starcoder>`__
       * `SantaCoder <https://huggingface.co/bigcode/gpt_bigcode-santacoder>`__
-  * - ``chatglm``
-    - ChatGLM
-
-      * Prebuilt not compiled yet
-      * `Official link <https://github.com/THUDM/ChatGLM-6B/blob/main/README_en.md>`__
-      * `Relax Code <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/chatglm.py>`__
+  * - `ChatGLM <https://github.com/THUDM/ChatGLM-6B/blob/main/README_en.md>`__
+    - * Prebuilt not compiled yet
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/chatglm.py>`__
     - 
     - * `ChatGLM2 <https://huggingface.co/THUDM/chatglm2-6b>`__
       * `CodeGeeX2 <https://huggingface.co/THUDM/codegeex2-6b>`__
+  * - `StableLM <https://huggingface.co/stabilityai>`__
+    - * Prebuilt not compiled yet
+      * `MLC Implementation <https://github.com/mlc-ai/mlc-llm/blob/main/mlc_llm/relax_model/stablelm_3b.py>`__
+    - 
+    - * `StableLM <https://huggingface.co/collections/stabilityai/stable-lm-650852cfd55dd4e15cdcb30a>`__
 
-If the model variant you are interested in is in one of these model architectures we support (but we have not provided the prebuilt weights yet), you can check the :doc:`model compilation page </compilation/compile_models>` on how to compile your own models. Note that you only need to compile the weights for your model variant and reuse the library file found in Level 2 tables.
+If the model variant you are interested in uses one of these model architectures we support (but we have not provided the prebuilt weights yet), you can check out :doc:`/compilation/compile_models` on how to compile your own models. Afterwards, you may follow :doc:`/compilation/distribute_compiled_models` to upload your prebuilt weights to hugging face, and submit a PR that adds an entry to this page, contributing to the community.
 
 For models structured in an architecture we have not supported yet, you could:
 
-- Either `create a new issue <https://github.com/mlc-ai/mlc-llm/issues/new/choose>`_ to request a new model architecture.
+- Either `create a [Model Request] issue <https://github.com/mlc-ai/mlc-llm/issues/new?assignees=&labels=new-models&projects=&template=model-request.md&title=%5BModel+Request%5D+>`__ which automatically shows up on our `Model Request Tracking Board <https://github.com/orgs/mlc-ai/projects/2>`__.
 
 - Or follow our tutorial :doc:`Define New Models </tutorials/customize/define_new_models>`, which introduces how to bring a new model architecture to MLC-LLM.
 
