CoruNethron

It's under Windows 7 x64, Google Chrome 63.0.3239.84 x64 Ubuntu 16.04 x64 - server side.

Seems like managed it to work, based on investigations from #607 Comment whole line: https://github.com/Kode/Kha/blob/c9c5ac70c4d77a50e49df78cc87584a590003afc/korefile.js#L33 Add to khafile.js in your project: ``` //... project.addParameter('-D HXCPP_TELEMETRY'); project.addParameter('-D HXCPP_STACK_TRACE'); project.addDefine('HXCPP_DEBUGGER'); project.addCDefine('HXCPP_DEBUGGER'); project.addDefine('HXCPP_PROFILER');...

Hello, thx for debug tips. Same issue as @verilog-indeed , unable to flush Tang Nano on **Windows 10**. openFPGALoader is running from commandline, so there is no /etc/udev/rules.d Running CMD...

Managed to workaround this, as mentioned here: https://esp32.com/viewtopic.php?t=15469 Using this tool: https://zadig.akeo.ie/  Replacing driver for Sipeed-JTAG from FTDIBUS to WinUSB seems to solve the issue. openFTDILoader starts to correctly...

Same here. Even with example curl request from the API docs page. Edited: I've noticed, that removing both `top_logprobs` and `logprobs` items completely from request solves the issue, unless you...

Managed to get deterministic output, calling ``` seed=123 random.seed(a=seed, version=2) torch.manual_seed(seed) np.random.seed(seed) ``` from Kandinsky2_1 __init__() I'll try to prepare some useful interface for it next few days.

@Blucknote I didn't try less calls yet, probably python''s random and numpy seeds is redudant in my solution. Also, maybe manual_seed_all() is not nrcessary in yours. Thank's. Also it may...

@Blucknote , I should mention, that i'm using an `mps` device during model initialization, not `cuda`, that's may make difference.

@wacdev Thank you for fixes. Tested on M1 16Gb, with `max_new_tokens=150, max_length=1000` and all other yours changes. It can be executed. Performs very slow, because of excessive swap usage. Description...

@simongcc please check this fork https://github.com/wacfork/MiniGPT-4/tree/main by @wacdev , there was few changes, like device type (if CUDA device is not available). As far as I remember, I just took...