bradhilton

6 ms·

1.
by bradhilton 10mo ago
The problem is that they have a lot of time to report their purchases. If they were required to report before they purchased the problem would probably resolve itself.
2.
by bradhilton 10mo ago
The SP500 is probably the most popular investment in America, perhaps aside from housing. Wouldn't hurt to have lawmaker's fortunes broadly aligned versus narrowly aligned with specific corporations.
3.
by bradhilton 1y ago
Awesome! If you run into any problems or have questions feel free to open an issue or drop by the discord [1] server. [1] https://discord.gg/zbBHRUpwf4
4.
by bradhilton 1y ago
Hi, we don't have reliable documentation for the HTTP API endpoints yet, mostly as they are still subject to change. However, to briefly provide some context, `/_train_model` returns a stream of line delimited JSON objects for eac
5.
by bradhilton 1y ago
Contributor here, we developed the Agent Reinforcement Trainer (ART) library to make it easy to train LLMs for anything. No callbacks or straitjacket flows. Instead we serve an OpenAI API-compatible endpoint that you can use as a drop-in re
6.
by bradhilton 1y ago
I could see training your own email agent being beneficial for products like this: https://x.com/advaitpaliwal/status/1913290027897131084
7.
by bradhilton 1y ago
I know Google DeepMind ran experiments with 10M a while ago, but I think this will be the first legit, released 10M context window model.
8.
by bradhilton 2y ago
Yes, pedantically, it is! But as I said, everything's on a spectrum. Online-ish data can still work just fine.
9.
by bradhilton 2y ago
We used about 58 hours on 4xH100s and about 19 hours on 8xH100s to get the very best result with the 32B model. We trained for about another 16 hours before finishing the run, but we could have stopped earlier after it was apparent the mode
10.
by bradhilton 2y ago
Well, in this case there is a much more straightforward method with the same CP-SAT solver used to create the puzzles. This is more of a fun experiment to see if we can train LLMs to solve these kinds of logical deduction problems.
11.
by bradhilton 2y ago
Technically yes, only if you do a gradient step with data sampled from the exact same weights is it an online step. With our training recipe this can be easily done by accumulating the gradients across the entire batch and only doing one st
12.
by bradhilton 2y ago
The model is rewarded for accuracy. For each puzzle there are a few multiple choice questions. If it got 1 out of 4 correct, for example, its reward would be 0.25. Then group relative advantages are calculated. If you have 16 different resp
13.
by bradhilton 2y ago
Yeah, the takeaway shouldn't be "our model is smarter," but that we were able to train weak models to as good or better than the best for this specific task. Depends on what you're doing, but sometimes that is enough.
14.
by bradhilton 2y ago
We updated the first paragraph to define the acronym. Thanks again for the feedback!
15.
by bradhilton 2y ago
Great question! So the dataset includes prompts and solutions, but no "gold" answer per se to use for SFT. You could sample responses from larger models and then train the smaller model on their answers, but as outlined in the ben
16.
by bradhilton 2y ago
Great point! Thanks for the feedback.
17.
by bradhilton 2y ago
Yeah, it may help. In this paper[1], the author used a KL penalty of 0.01 for general tasks and 0.001 for mathematical. I tend to think it's probably not very important unless you're trying to optimize for human preferences. As fo
18.
by bradhilton 2y ago
We trained all the parameters. Those would definitely be interesting ablations. I would also like to see how much of a performance hit we would take with PEFT methods like LoRA.
19.
by bradhilton 2y ago
No meaningful changes to the hyperparameters, just changed the tasks per iteration to 16 and trained on the same first 16 training tasks each iteration. We only tested this with the 14B model. You can see the run here: https://wa
20.
by bradhilton 2y ago
As for why they dropped suddenly , I don't really know. Sometimes models develop degenerate behaviors, but even when forking from the best checkpoint and lowering the learning rate or changing other hyperparameters, performance stills
21.
by bradhilton 2y ago
Hi, other author here. I think the models converged on shallow/greedy strategies that improved performance up to a point, but are ultimately shortsighted, especially for harder puzzles. Something interesting I noticed in the responses
22.
by bradhilton 2y ago
They need to be trained with a small length penalty
23.
by bradhilton 2y ago
These are distillation fine-tunes of two different models: - Qwen2.5 7B - Llama3.1 8B Though the sizes are similar, they will probably have different strengths and weaknesses based on their lineage.
24.
by bradhilton 2y ago
Beats Gemini 1.5 Pro at all but two of the listed benchmarks. Google DeepMind is starting to get their bearings in the LLM era. These are the minds behind AlphaGo/Zero/Fold. They control their own hardware destiny with TPUs. Bulli
25.
by bradhilton 2y ago
Okay, that was really impressive. Well done!
26.
by bradhilton 2y ago
Kind of feels like the ride-sharing early days. Lots of capital being plowed into a handful of companies to grab market share. Economics don't really make sense in the short term because the vast majority of cash flows are still far in
27.
by bradhilton 2y ago
I'm glad he vetoed the bill, but his rationale is worrisome. Even if he's just trying to placate SB 1047 proponents, they will try to exact concessions from him in future sessions. I'll take this brief reprieve, but it's
28.
by bradhilton 2y ago
Some parents try to have their cake and eat it too by altering the spelling or pronunciation of otherwise common names, thus ensuring their child both fits in and is unique. Cheekiness aside, naming our children has been a fun, stressful,
29.
by bradhilton 3y ago
Absolutely, you should look at XGBoost, LightGBM, and CatBoost. XGBoost is the og and the most feature-rich. LightGBM is the fastest and what I use for my case (millions of rows of data with over 100 features). CatBoost could be good depend
30.
by bradhilton 3y ago
What's in common with those industries is that they are largely regulated or government-controlled, making them much more difficult to break into. Don't hate the players, hate the game. If we deregulated there would be a rush of i

More ›