This is just someone's personal blog/opinion. I wouldn't read too much into it... "The site is run by Zygmunt Zajc (pronounced “Ziontz”). ... An economist by education"
I flagged it for these reasons as well. It's just a bad article. Shows very poor understanding of the basics of LLM workings, and the field in general.
Lingers on the "cheated" benchmark (lmsys) but never mentions all the other 3rd party benchmarks performed after the inference fixes, which are in line with what Meta originally published. To be clear, submitting a different fine-tuned model to one arena and releasing the untuned model without clearly mentioning this, is bad. But conflating the "human prefference" bench with the others and not mentioning the models capabilities on other benchmarks is also bad writing.
The MoE paragraphs are bad, and the writer never explains why the copy 17B vs VRAM size is bad, they just leave it there unexplained.
Poor form, I was expecting better from someone working in this field.
> OpenAI would not disclose exactly how large GPT-4o mini is, but said it’s roughly in the same tier as other small AI models, such as Llama 3 8b, Claude Haiku and Gemini 1.5 Flash.
As far as I can tell all of the 8B rumors were seeded by that loosely sourced comparison to Llama 3 8B.
I know for a fact that Gemini 1.5 Flash is NOT an 8B model, because a separate model called "Gemini 1.5 Flash 8B" came out after that article was published - the first Gemini model with a documented parameter count. Flash 8B is priced at half the cost of regular Flash.
There's also this paper that mentions 8B but provides no source for that at all, which makes me suspect their initial source may have been that TechCrunch rumor: https://arxiv.org/pdf/2412.19260
we are in for a lot of pain if seemingly intelligent people make mistakes like this. grabbing the number of params from what gpt gives you. how can you do that?
The comment at the top says it's a draft. It's not unreasonable to ask for random values from a GPT for "filler" for the draft (or even just make them up), just to stay in the flow, and then track down the real numbers later.
Do you think there are more bugs in Llama 4 at this time? Or have the bugs been patched, and the current version of llama.cpp + whatever the latest GGUF version is would be representative of the true performance of Llama 4?
I see you've uploaded new Maverick GGUF/safetensors files yesterday, along with a lot of other models like Deepseek R1, was there an issue with the older model files?
For English, on a combination of MixEval, LiveBench, IFEval, and EvalPlus Maverick FP8 (17B/400B) was about on par with DeepSeek V3 FP8 (37B/671B) and Scout (17B/109B) was punching in the ballpark of Gemma 3 27B, but not too far off Llama 3.3 70B and Mistral Large 2411 (123B).
Llama 4 claimed to be trained on 10X more multilingual tokens than Llama 3 and testing on Japanese (including with some new, currently unreleased evals) the models did perform better than Llama 3 (although I'd characterize their overall Japanese performance as "middle of the pack": https://shisa.ai/posts/llama4-japanese-performance/
I think a big part of the negative reaction is that in terms of memory footprint, Llama 4 looks more built for Meta (large scale inference provider) than home users, although with the move to APUs and more efficient CPU offloading, there's still something to be said for strong capabilities at 17B of inference.
I think people are quick to forget that Llama 3, while not so disastrous, was much improved with 3.1. Also the competitive landscape is pretty different now. And I think the visual capabilities are being a bit slept upon, but I think that's also the case of releasing before the inference code was baked...
This seems to be a general problem at the moment. The most usable models are not the newest. The newer models (obviously, I haven't tried them all) may do better on benchmarks, but actual usability is worse.
To create useful LLMs required some genuine breakthroughs. It seems to me that we have reached the limits of what we can achieve with current architectures. Progress will require some new insights and breakthroughs.
If you game the benchmark then you always get found out by your users. Yet the practice remains common in hardware. Outright lies are uncommon but misleading and cherry picked numbers are pretty much standard practice.
The fact that misleading benchmarks don't even drive profit at Meta didn't seem to stop them doing the same thing, but perhaps this isn't very surprising. I imagine internal incentives are very similar.
Unlike the hardware companies though, gaming the benchmark in LLMs seems to involve making the actual performance worse, so perhaps there is more hope that the practice will fade away in this market.
Spock: Logic is a little tweeting bird chirping in a meadow. Logic is a wreath of pretty flowers which smell bad. Are you sure your circuits are registering correctly? Your ears are green.
Readit News
"just in front of GPT-4o-mini, which is, according to itself, a model with 1.3B or 1.5B or 1.7B parameters, depending on when you ask."
Then later:
"On the Artificial Analysis benchmark Scout achieved the same score as GPT 4o mini. A 109B model vs a 1.5B model (allegedly). This is ABYSMAL."
Asking models how many parameters they have doesn't make sense.
There is absolutely no way GPT-4o mini is 1.5B. I can run a 3B model on my iPhone, but it's a fraction of the utility of GPT-4o mini.
Lingers on the "cheated" benchmark (lmsys) but never mentions all the other 3rd party benchmarks performed after the inference fixes, which are in line with what Meta originally published. To be clear, submitting a different fine-tuned model to one arena and releasing the untuned model without clearly mentioning this, is bad. But conflating the "human prefference" bench with the others and not mentioning the models capabilities on other benchmarks is also bad writing.
The MoE paragraphs are bad, and the writer never explains why the copy 17B vs VRAM size is bad, they just leave it there unexplained.
Poor form, I was expecting better from someone working in this field.
GPT-4o mini is supposed to be ~8b params from estimates.
> OpenAI would not disclose exactly how large GPT-4o mini is, but said it’s roughly in the same tier as other small AI models, such as Llama 3 8b, Claude Haiku and Gemini 1.5 Flash.
As far as I can tell all of the 8B rumors were seeded by that loosely sourced comparison to Llama 3 8B.
I know for a fact that Gemini 1.5 Flash is NOT an 8B model, because a separate model called "Gemini 1.5 Flash 8B" came out after that article was published - the first Gemini model with a documented parameter count. Flash 8B is priced at half the cost of regular Flash.
There's also this paper that mentions 8B but provides no source for that at all, which makes me suspect their initial source may have been that TechCrunch rumor: https://arxiv.org/pdf/2412.19260
1. RMS norm eps was 1e-6, but should be 1e-5 - see https://github.com/huggingface/transformers/pull/37418
2. Llama 4 Scout changed RoPE settings after release - conversion script for llama.cpp had to be fixed. See https://github.com/ggml-org/llama.cpp/pull/12889
3. vLLM and the Llama 4 team found QK Norm was normalizing across entire Q & K which was wrong - accuracy increased by 2%. See https://github.com/vllm-project/vllm/pull/16311
If you see https://x.com/WolframRvnwlf/status/1909735579564331016 - the GGUFs I uploaded for Scout actually did better than inference providers by +~5% on MMLU Pro. https://docs.unsloth.ai/basics/tutorial-how-to-run-and-fine-... has more details
I see you've uploaded new Maverick GGUF/safetensors files yesterday, along with a lot of other models like Deepseek R1, was there an issue with the older model files?
For English, on a combination of MixEval, LiveBench, IFEval, and EvalPlus Maverick FP8 (17B/400B) was about on par with DeepSeek V3 FP8 (37B/671B) and Scout (17B/109B) was punching in the ballpark of Gemma 3 27B, but not too far off Llama 3.3 70B and Mistral Large 2411 (123B).
Llama 4 claimed to be trained on 10X more multilingual tokens than Llama 3 and testing on Japanese (including with some new, currently unreleased evals) the models did perform better than Llama 3 (although I'd characterize their overall Japanese performance as "middle of the pack": https://shisa.ai/posts/llama4-japanese-performance/
I think a big part of the negative reaction is that in terms of memory footprint, Llama 4 looks more built for Meta (large scale inference provider) than home users, although with the move to APUs and more efficient CPU offloading, there's still something to be said for strong capabilities at 17B of inference.
I think people are quick to forget that Llama 3, while not so disastrous, was much improved with 3.1. Also the competitive landscape is pretty different now. And I think the visual capabilities are being a bit slept upon, but I think that's also the case of releasing before the inference code was baked...
I have no idea how the author can remotely trust GPT-4o-mini in this case. The number of parameters is almost certainly way off.
To create useful LLMs required some genuine breakthroughs. It seems to me that we have reached the limits of what we can achieve with current architectures. Progress will require some new insights and breakthroughs.
The fact that misleading benchmarks don't even drive profit at Meta didn't seem to stop them doing the same thing, but perhaps this isn't very surprising. I imagine internal incentives are very similar.
Unlike the hardware companies though, gaming the benchmark in LLMs seems to involve making the actual performance worse, so perhaps there is more hope that the practice will fade away in this market.
Spock: Logic is a little tweeting bird chirping in a meadow. Logic is a wreath of pretty flowers which smell bad. Are you sure your circuits are registering correctly? Your ears are green.
https://www.imdb.com/title/tt0708432/quotes/?item=qt0406609