ncoder
ncoder
Still not fixed? My error message looks the same.
> Thanks for the solution, now setup-cuda.py works, but when I try to load model I get this error: > > ``` > Loading llama-7b-hf... > Traceback (most recent call...
FYI, I've also managed to get it to work with the new version of GPTQ, but i had to re-quantize the weights. On Tue, Mar 21, 2023 at 7:43 AM...
I also reproduce this bug. Looks like there are extra /@fs tags in front of the module path during registration that are missing when the reload event happens.
```txt async function fetchUpdate(runner, { path, acceptedPath }) { const maps = getCache(runner); - const mod = maps.hotModulesMap.get(path); + var mod = maps.hotModulesMap.get(path); if (!mod) { - return; + mod...
I also reproduce this issue. On macos. Debug log shows this is the cause ```txt GET https://api.github.com/repos/[redacted]/actions/secrets/public-key 404 ```
Here's a bit more about the difficulty of the topic: https://stackoverflow.com/questions/60842629/how-to-serialize-64-bit-integer-in-javascript
About a). Looks like using gRPC would work on nodejs would work. https://grpc.io/docs/languages/node/basics/
> Alternatively, you could also use UUID, there is a direct conversion between i64 and uuid. I don't understand what you mean. UUIDs are 128 bit values in my book.
Do you mean to say, that for a point with id 6763074405716742595 for example , i should fetch with uuid: 00000000000000005ddb4456f64851c3 ? By that i mean zero-pad it? The interface...