Viewing profile — gpugreg

Sure! But where?

To learn about sentiment analysis, I'd look for related datasets and then look at recent code, e.g. here: https://www.kaggle.com/datasets?search=sentiment+analysis For more LLM-spe…

Agents usually start with ingesting the existing code base, and DeepSeek can use those code bases for pretraining. And they will have filters on top of that to throw out garbage. I…

Oh, I messed up. Half-way through, I thought it would be a good idea to double the numbers so I don't have to deal with half millions, but forgot to also double the 98.5. Unfortuna…

Agentic workloads are somewhere around 1%/0.5%/98.5% input/output/cached tokens. Cached tokens are pretty much free for inference providers (if they implement sparse and compressed…

Personally, I prefer QDirStat. I just tried to use FileLight to compare, but the package seems to be broken on Lubuntu.

> you're having issues handling files properly? I guess they were using ollama, which does not tell you where it puts the models it downloads.

I get the following error: Traceback (most recent call last): File "/app.py", line 1, in import spider ModuleNotFoundError: No module named 'spider' Steps to reproduce: 1. Visit ht…

I did not say that it is impossible. I just think that we need architectural improvements, or maybe even a fundamentally different approach to get something like Kimi K3 for cheap.…

Looks like global energy consumption has risen by an order of magnitude from 1900 to 2000: https://www.encyclopedie-energie.org/en/world-energy-consump... Unfortunately, electricit…

> Some of us might be rich I sure wish I had a few 100M of disposable income to train a frontier model. > or in the future it could be useful when training is cheaper. I do not thi…

Where do you see less than $10/h for 8 * MI354X? I can only find $2.50 for 1 * MI355X (lowest I can find for rent on other websites is $2.65, but maybe they got a better deal).

Notably, MXFP4 was introduced at the (much less costly) supervised fine-tuning stage after pretraining, so the number of B200/B300 GPUs could be relatively small in comparison to t…

> they also mentioned only having a 20K GPU cluster (unclear if NVIDIA, or Huawei). A few quotes from the transcript: > Our current computing capacity is approximately 20,000 H-equ…

dax (coauthor) recently tweeted https://xcancel.com/thdxr/status/2083178051052155182 > because we added the new deepseek which we do not yet have a ZDR with we cannot blanket say w…

The memory bandwidth of the 2x RTX Pro 6000 Blackwell setup will be 10x higher, which should have an equivalent effect on the generated tokens per second.

Unfortunately, all mentions of ZDR have silently been removed from the OpenCode Go page today.

To add to this, the $60 only applies to DeepSeek-V4-Flash and a few other models. For DeepSeek-V4-Pro, the amount is $15. https://opencode.ai/docs/go/#usage-limits Previously, Open…

According to the leaked call transcript, DeepSeek is working on vision for V4. Not sure when it will land though.

I think this is less about winning the AGI race and more about not dropping out of AI entirely.

The model is already natively MXFP4-quantized during training, so there is no quality loss.

Brussels is set to contribute roughly €5 billion, matched by another €5 billion from European governments, alongside around €20 billion in private investment. You can train a model…

From the Kimi K3 technical report: Kimi K3 supports a context window of up to 1 million tokens. We achieve this through extending the context window progressively as training proce…

Anthropic probably trained Mythos on their own code and found that it is too got at reproducing it.

Some suits with no understanding of how LLMs work are scared that the models might hack them, or believe that they'd have to send data to China because they do not know that open m…