aubanel

12 ms·

121.
A brief history of AI from the 50s to today's LLMs – without any math(aymeric-roucher.github.io)
1 points by aubanel 2y ago|0 comments
122.
by aubanel 2y ago
Impressive to see the positivity that both him and his wife Bess maintained in their last posts, despite the constant pain and oncoming death. Real strength. Rest in Peace Jake, and I wish you the best on the path forward Bess!
123.
by aubanel 2y ago
Great, now I can finally find the top 5 hiding places for these friggin campers!
124.
AI Lab Kyutai's groundbreaking model Moshi: conversations with <200ms latency(youtube.com)
2 points by aubanel 2y ago|0 comments
125.
New LLM Agent writing actions in Python code tops the GAIA agent benchmark(huggingface.co)
2 points by aubanel 2y ago|0 comments
126.
by aubanel 2y ago
Just to be sure: the post does not say that the performance of Open LLMs is plateauing (because that would be false, e.g. Google just released a Gemma2 that blows out of the water all previous open models from the same sizes). Its true titl
127.
by aubanel 2y ago
LMSys Chatbot Arena is a crowd-sourced ranking with an ELO system: basically users a presented with 2 hidden models, they get the answers of the 2 models when presenting their request, and they vote which one performed bests, which realized
128.
by aubanel 2y ago
It's exceptionally strong. In LMSys Chatbot Arena, the 27B version scores above LLama-3-70B, at the level of OpenAI GPT-4 and Claude-3 Sonnet!
129.
by aubanel 2y ago
Something great I've specifically witnessed yesterday: when doing a small request, like ask for a feature request in the team channel, don't just describe a small use case. Instead, provide the bigger picture: e.g. "I'm
130.
by aubanel 2y ago
This model has: 1. On par or better performance than Llama-3-70B-Instruct 2. A much more comfortable context length of 128k (vs the tiny 8k that really hinders Llama-3) These 2 feats together will probably make it the first serious OS rival
131.
by aubanel 2y ago
Previous post about the version 1 of Agents: https://news.ycombinator.com/item?id=35889743 We've done a complete revamp of the package since then, with two goals in mind: - clarity through simplicity: as few abstractio
132.
Hugging Face launches Agents 2.0(huggingface.co)
7 points by aubanel 2y ago|1 comments
133.
by aubanel 2y ago
Please elaborate on this: how is it flawed?
134.
by aubanel 2y ago
> Additionally, members of the program receive priority placement and “richer brand expression” in chat conversations This sounds particularly bad since it's the polar opposite of what Sam Altman himself pretended to want in his rec
135.
by aubanel 2y ago
Wow that's disgusting. We generally agree after some criminal decisions that decision-makers should be punished, and political parties should be forbidden. Why not apply this to companies ? I see the benefits (punishing stakeholders an
136.
by aubanel 2y ago
Putting "cut-throat bureaucrats" in the "do not care for impact" side of the axis seems unnatural to me: I think these people do care for impact, and that's why they are so decided about imposing their ways. But the
137.
by aubanel 2y ago
On WizardLM2: the weights have been pulled back by Microsoft from HF Hub, as there seems to be internal debate on whether to really publish it. There's a risk that it does not ever get published again. So keep your local copies if you
138.
by aubanel 3y ago
The calls with long context are done through specific APIs, that you can call for instance in Python or Javascript. Here's a quick start guide with OpenAI: https://platform.openai.com/docs/quickstart?context=python
139.
Show HN: AI Travel Planner Demo(huggingface.co)
1 points by aubanel 3y ago|0 comments
140.
by aubanel 3y ago
> I think in summary, a better approach would’ve been “You obviously know the 50 states, GPT, so just give me the full name of the state this pertains to, or Federal if this pertains to the US government.” Why not really compare the two
141.
by aubanel 3y ago
I think agents are not a fad, they're here to stay since using an LLM in an agent system is the only way to let it access real world task, which they will end up doing when they're good and reliable enough. That said, I believe th
142.
by aubanel 3y ago
Even in the most restrictive définition of recall as in "retrieve a short contiguous piece of information inside an unrelated context", it's not that good. It's always the exact same needle inserted in the exact same con
143.
by aubanel 3y ago
Be aware the haystack test is not good at all (in its current form). It's a single piece of information inserted in the same text each time, a very poor measurement of how well the LLM can retrieve info.
144.
by aubanel 3y ago
I love the example with Russian elections! It's impressive that they allow themselves to be that low-effort in their fake results.
145.
by aubanel 3y ago
Did learning about the brain being made of meat also destroy your awe of human intelligence? I feel like being able to look inside the clockwork should not distract us from being amazed at how wonderful a system can be.
146.
by aubanel 3y ago
The Rafale is cheaper than the F35, by a wide margin.
147.
Show HN: Visualize how you split your document into chunks for RAG applications(huggingface.co)
2 points by aubanel 3y ago|0 comments
148.
by aubanel 3y ago
> My New Year's Resolution in 2019 was to learn the piano. Key to success is building mental representations of music and the keyboard. Increasingly, I do this by playing pieces on my mental piano As a correct pianist (lots of class
149.
by aubanel 3y ago
Systematically calling russian people Orcs is the worst dehumanising things to do IMO. I have nothing to do with that conflict but I don't like how confident some people are that they're on the side of the Good, thus that they cou
150.
by aubanel 3y ago
I prefer my theory, it's funnier although less probable, I agree

More ›