CoderSupreme

CogVLM: Visual Expert for Pretrained Language Models

Presents CogVLM, a powerful open-source visual language foundation model that achieves SotA perf on 10 classic cross-modal benchmarks

repo: https://github.com/THUDM/CogVLM abs: https://arxiv.org/abs/2311.03079

article: https://x.ai

trained a prototype LLM (Grok-0) with 33 billion parameters. This early model approaches LLaMA 2 (70B) capabilities on standard LM benchmarks but uses only half of its training resources. In the last two months, we have made significant improvements in reasoning and coding capabilities leading up to Grok-1, a state-of-the-art language model that is significantly more powerful, achieving 63.2% on the HumanEval coding task and 73% on MMLU.

I just came across a very useful tip for those of you who are looking to migrate your VSCode extensions to VSCodium. Instead of removing VSCode altogether, you can simply link the folders and keep both editors installed on your system.

Here's how to do it:

1. Install VSCode and all the extensions you want to migrate.
2. Copy the extensions folder from VSCode. You can find it by going to the VSCode settings and searching for "extensions".
3. Install VSCodium on your system.
4. Move the extensions folder you copied earlier to the VSCodium extensions folder. You can find it by going to the VSCodium settings and searching for "extensions".
5. Finally, you can remove VSCode if you no longer need it.

By following these steps, you can easily migrate your VSCode extensions to VSCodium without losing any of your settings or configurations. And if you ever want to switch back to VSCode, you can simply unlink the folders and continue using it.

I hope you find this tip helpful. Let me know if you have any other tips in the comments below.

To filter out high-pitched sounds on Linux using PulseAudio, you can use a low-pass filter. A low-pass filter allows sounds below a certain frequency to pass through while filtering out higher frequencies.

For MPV, you can apply a low-pass filter directly when playing a media file. The command would look like this:

mpv --af=lowpass=f=3200 yourfile.mp4

In this command, f=3200 sets the cut-off frequency to 3200 Hz. All frequencies above this will be filtered out.

For a system-wide solution, you can configure PulseAudio to apply a LADSPA low-pass filter to all audio. This process is more complex and involves installing and configuring the LADSPA plugin. The exact steps can vary depending on your system configuration and the specific LADSPA plugin you choose to use.

Another option is to use PulseEffects, a powerful real-time audio processing tool that works with PulseAudio. PulseEffects allows you to apply a variety of filters and effects to your audio, including a low-pass filter. You can adjust the cut-off frequency of the filter to target the high-pitched sounds you want to filter out.

Remember that filtering out high frequencies can affect the overall sound quality, potentially making it sound muffled or dull. You may need to experiment with different cut-off frequencies to find a balance that works for you.

I'm trying to set up a producer/consumer pipeline using asyncio in Python. The goal is to have one coroutine that reads input continuously and puts it into a queue, while another coroutine pulls items from the queue and processes them in the background.

I think I need to use asyncio queues and tasks, but not quite sure how to structure it properly. Could someone provide a simple asyncio producer/consumer queue example? Ideally it would show both coroutines, use asyncio.gather to run them concurrently, and handle the event loop setup/cleanup.

Any help would be really appreciated!