what llama-chat produces:

What it should look like (on the example of chatgpt)

Hey!

I am getting an error while getting data from llama-3-8b* and llama-3-70b*.

above you can see the screenshot of the error.

Thanks!

Is Synchronous Non-Stream API supported or how can I modify the code to adopt a non-stream API?

A second file can be uploaded, but Llava and Salmonn keep referring to the previously uploaded file. I believe this is because they get sent the messageHistory, and this is prioritised over the file.

To replicate:

1. Upload an image or audio file.
2. Ask "What's in this [image/audio file]?" and receive description of file 1.
3. Upload a second image or audio file.
4. Ask "What's in this [image/audio file]?" and receive description of file 1 again (or even "In the audio file, there is a person speaking about the [contents of image 1]" if you switched between an image and an audio file).

A simple fix would be to (warn the user and) remove all previous message history when a new file is uploaded.

Syntex is not highlighted or Programming is not colorized that any type of return token from LLama

Sample chat code works fine locally; getting this error when deployed to Vercel:

llama-98079bc32b437141.js:1 Uncaught (in promise) Error: Server error: A server error has occurred

FUNCTION_INVOCATION_FAILED

at llama-98079bc32b437141.js:1:7882
at l (main-1a90909837c5cc42.js:1:2020)
at Generator._invoke (main-1a90909837c5cc42.js:1:1808)
at P.forEach.e.<computed> [as next] (main-1a90909837c5cc42.js:1:2443)
at v (llama-98079bc32b437141.js:1:4641)
at o (llama-98079bc32b437141.js:1:4844)

As an optimization, we can cancel any running predictions when the user closes the page.

Unhandled Runtime Error
Error: Objects are not valid as a React child (found: [object Error]). If you meant to render a collection of children, use an array instead.

Call Stack
throwOnInvalidObjectType
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (8872:0)
reconcileChildFibersImpl
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (9879:0)
reconcileChildFibers
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (9900:0)
reconcileChildren
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (15606:0)
updateHostComponent$1
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (16568:0)
beginWork$1
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (18390:0)
beginWork
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (26741:0)
performUnitOfWork
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (25587:0)
workLoopSync
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (25303:0)
renderRootSync
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (25258:0)
performSyncWorkOnRoot
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (24727:0)
flushSyncWorkAcrossRoots_impl
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (10274:0)
flushSyncWorkOnAllRoots
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (10234:0)
processRootScheduleInMicrotask
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (10379:0)
eval
node_modules\next\dist\compiled\react-dom\cjs\react-dom.development.js (10550:0)

When selecting Llava from drop down. Any filetype attempted in image upload fails. Error: does not accept file type jpeg, png etc. Latest version fetched today.

When i send a message to the ChatBot i get this Message "Objects Are Not Valid as a React Child"

Please fix it

In Replicate, the Llama-3-8b-instruct model takes in a "system_prompt" and "prompt_template" param. The replicate docs mentioned that system_prompt is prepended to the text being sent, and I'm assuming the prompt_template is also used to format the prompt data since there's a default value that includes the special tokens. Is there a reason this application is formatting the prompt to also use these special tokens? I guess I'm just a little confused how this is being handled by Replicate behind the scenes.

For example, for the message "Can you tell me about the story of David and Goliath" the front-end sends:

{
  system_prompt: "You are a helpful assistant.",
  prompt: "<|begin_of_text|><|start_header_id|>system<|end_header_id|>
You are a helpful assistant.<|eot_id|>
<|start_header_id|>user<|end_header_id|>

Can you tell me about the story of David and Goliath<|eot_id|>"
}

The default "prompt_template" seems to be:

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>

{prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>

So, behind the scenes, it seems like what gets passed to the model becomes:

<|begin_of_text|><|start_header_id|>system<|end_header_id|>

You are a helpful assistant.<|eot_id|><|start_header_id|>user<|end_header_id|>

<|begin_of_text|><|start_header_id|>system<|end_header_id|>
You are a helpful assistant.<|eot_id|>
<|start_header_id|>user<|end_header_id|>

Can you tell me about the story of David and Goliath<|eot_id|><|eot_id|><|start_header_id|>assistant<|end_header_id|>

since the prompt and system_prompt are possibly directly passed into the prompt template. Is this a correct understanding?

After uploading the image and giving a prompt to it . The Model stops working and throws an application error

LLAVA not working

Currently seeing this in macOS Chrome at https://llama.replicate.dev

@replicate/hackers

On opening https://llama.replicate.dev/ when i start a chat with ai press enter a window pops up which shows Application error: a client-side exception has occurred (see the browser console for more information).

Using an input all linebreaks are lost. If I copy some text into it and ask llama2 something about that text, all paragraphs are lost, which might change llama2's output.

This is especially problematic when coding with llama2 as an assistant.

Cloudflare Captcha says : Invalid domain (https://www.llama2.ai/) Contact the site admin if this problem persists

#71 added performance metrics, which display above the message form. However, the last message of conversations that span beyond the viewport height are now cut off.

Since the text containing div uses flex the text is displayed like it had a linebreak after each span. It looks properly, however, once text is copied it has no linebreaks (as the text in the HTML has none)

This makes copying generated texts, lists, code difficult, as all visible text lines are merged into one mess without linebreaks.

I foundt it very careful for answers, so I wanna know it's 70B or 70B-chat?

Seeing this error in the console:

{
  "detail": "CUDA error: an illegal memory access was encountered\\nCUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.\\nFor debugging consider passing CUDA_LAUNCH_BLOCKING=1.\\nCompile with `TORCH_USE_CUDA_DSA` to enable device-side assertions."
}

Related to #71.

The blank state for the total time metric displays a zero value instead of a placeholder dash like the others. It should be consistent and display as "—" until a value is set.

When click the "Chat" button, the interface will display the following content:

Application error: a client-side exception has occurred (see the browser console for more information).