-
Delta Chat is a protocol that uses existing email infrastructure (IMAP / SMTP) to deliver encrypted messages: https://delta.chat/en/ - the "global identity" would definitely still be red for Delta Cha…
-
The snort rules: /etc/snort/rules are circa 2015
**websnort** does not indicate how 'old' the snort rules are that a user is relying on for detection, I realize you do not develop this tool, etc. s…
-
```
trainer = ORPOTrainer(
model=model,
train_dataset=dataset["train"],
eval_dataset=dataset["test"],
#peft_config=peft_config,
tokenizer=tokeni…
-
Since the Raspberry Pi Foundation released the Pico SDK 1.5.0 with Bluetooth support, do you plan on releasing Circuitpython with Bluetooth support for the pico?
-
### Feature request
Is there a way to make the NCCL timeout configurable as we often get timeout problems with the starcoder model?
`[E ProcessGroupNCCL.cpp:828] [Rank 1] Watchdog caught collective …
-
There are already many projects that have attempted to create a distributed peer-to-peer social network, for example [diaspora*](http://diasporafoundation.org) and [RetroShare](https://retroshare.gith…
-
# Bug Report
## Description
edit: Even more people have experienced it now. It seems like sending message works all the time, but receiving never works. Receiving from store nodes works however.…
-
### Is there an existing issue for this?
- [X] I have searched the existing issues
### Current behaviour
The GetBlockSubsidy function could potentially encounter a signed integer overflow if …
-
### Your current environment
docker 0.4.0.post1
### 🐛 Describe the bug
```
docker run -d --runtime=nvidia --gpus '"device=1"' --shm-size=10.24gb -p 5001:5001 -e NCCL_IGNORE…
-
I tried this `NCCL_IGNORE_DISABLED_P2P=1 CUDA_VISIBLE_DEVICES=1,5,6,7 python vllm_inference.py --model_name model_hub/llama2-7B-chat-hf --tp_size 4`
and found that the memory usage on GPUs was like t…