mlab

I can mention that we have temporarily solved this by patching (using patch-package) the node module for this library. Its not a pretty patch but seems to work.

Had the same issue. Was able to access the emit method by attaching the context to the component: `SomeComponent.contextType = SocketContext;` And then in the component access emit via: `this.context.emit...`...

> @smith-co @thisisanshgupta @tlkh > > For torch, I wrote up a minimal example in deepspeed, which can train the 16B on a ~24 GB gpu. You would need to...

I had the exact same issue. Traced it to a request sent in the httpx library. Seems like the response given is 400 Bad Request, but this is somewhere incorrectly...

I am getting the following error when attempting to fine-tune: Traceback (most recent call last): File "/opt/gpt-j-8bit/gpt-j-6b-8-bit.py", line 242, in out = gpt.forward(**batch,) File "/opt/gpt-j-8bit/.env/lib/python3.8/site-packages/transformers/models/gptj/modeling_gptj.py", line 782, in forward transformer_outputs...