red2awn

8 ms·

1.
by red2awn 4mo ago
I have exactly the same thought. Anyone who had used seedance 2.0 a bit can tell Gemini is a bit behind, and seedance 2.1 is on the horizontal already.
2.
by red2awn 7mo ago
LLM-as-a-judge is quite effective method to RL a model, similar to RLHF but more objective and scalable. But yes, anthropic is making it more serious than it is. Plus DeepSeek only did it for 125k requests, significantly less than the other
3.
by red2awn 7mo ago
The "distillation attacks" are mostly using Claude as LLM-as-a-judge. They are not training on the reasoning chains in a SFT fashion.
4.
by red2awn 7mo ago
What's the SOTA open source or weight available turn taking model these days? I tried pipecat/smart-turn-v3 and the results are not good. It only works well when you say a short sentence in a clear voice. Anything else will cause
5.
by red2awn 7mo ago
Distilling from a teacher (Opus 4.5) and scaling RL more.
6.
by red2awn 7mo ago
> i need to wash my helicopter at the helicopter wash. it is 50m away, should i walk or fly there with my helicopter. Sonnet 4.6: Walk! Flying a helicopter 50 metres would be more trouble than it's worth — by the time you've do
7.
What nobody teaches you after you stop being broke(twitter.com)
2 points by red2awn 8mo ago|0 comments
8.
by red2awn 8mo ago
It was mentioned that the performance of DuckDB is similar to that of Polars (among others). In that case why would one choose DuckDB over Polars? The only differentiator seems to be that you do the querying with standard SQL instead of the
9.
by red2awn 8mo ago
They are heavily focused on code. Claude Code likely generates 100 of millions lines of Python a day, make the language a little bit better with $1.5M is extremely high leverage.
10.
by red2awn 9mo ago
Implement transformers yourself (ie in Numpy). You'll never truly understand it by just watching videos.
11.
by red2awn 9mo ago
It does NOT. Response healing only fixes JSON syntax errors, not schema differences. See https://news.ycombinator.com/item?id=46332119
12.
by red2awn 9mo ago
Tokens are sampled from logits using the constraints after a normal forward pass. The forward pass is the expensive part of LLM inference which isn't affected by structured output.
13.
by red2awn 9mo ago
Cool stuff! I don't get how all the open source inference framework have this down but the big labs doesn't... Gemini [0] is falsely advertising this: > This capability guarantees predictable and parsable results, ensures forma
14.
by red2awn 9mo ago
Very confused. When you enable structured output the response should adhere to the JSON schema EXACTLY, not best effort, by constraining the output via guided decoding. This is even documented in OpenRouter's structured output doc >
15.
by red2awn 9mo ago
Very interesting release: * Hybrid MoE: 2-3x faster than pure MoE transformers * 1M context length * Trained on NVFP4 * Open Source! Pretraining, mid-training, SFT and RL dataset released (SFT HF link is 404...) * Open model training recipe
16.
by red2awn 9mo ago
Opus 4.5 is the highest quality code I've seen out of LLMs, still some way to go to match programmers who care, but much better than most people. I find it enough to let it write the code and then manually polish it afterwards.
17.
by red2awn 9mo ago
Claude Code does not support AGENTS.md, you can symlink it to CLAUDE.md to workaround it. Anthropic: pls support!
18.
by red2awn 9mo ago
Correct, it's breaks the single prompt, single completion assumption baked into the frameworks. Conceptually it's still prompt/completion but for low latency response you have to do streaming KV cache prefill with a websocket
19.
by red2awn 9mo ago
Nice work. Are you working on streaming input/output?
20.
by red2awn 9mo ago
Why would you use an Omni model for text only workload... There is Qwen3-30B-A3B.
21.
by red2awn 9mo ago
They had a Flash variant released alongside the original open weight release. It is also mentioned in Section 5 of the paper: https://arxiv.org/pdf/2509.17765 For the evals it's probably just trained on a lot of t
22.
by red2awn 9mo ago
It is an in-house closed weight model for their own chat platform, mentioned in Section 5 of the original paper: https://arxiv.org/pdf/2509.17765 I've seen it in their online materials too but can't seem to f
23.
by red2awn 9mo ago
Flash is a closed weight version of https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct (it is 30B but with addtional training on top of the open weight release). They deploy the flash version on Qwen's own ch
24.
by red2awn 9mo ago
None of inference frameworks (vLLM/SGLang) supports the full model, let alone non-nvidia.
25.
by red2awn 9mo ago
The previous -Flash weight is closed source. They do have weights for the original model that is slightly behind in performance https://huggingface.co/Qwen/Qwen3-Omni-30B-A3B-Instruct
26.
by red2awn 9mo ago
This is a stack of models: - 650M Audio Encoder - 540M Vision Encoder - 30B-A3B LLM - 3B-A0.3B Audio LLM - 80M Transformer/200M ConvNet audio token to waveform This is a closed source weight update to their Qwen3-Omni model. They had a
27.
by red2awn 10mo ago
No, this is used for model alignment during post-training, not part of the system prompt. Why this is in the training data such that Claude can regurgitate it is currently unclear.
28.
by red2awn 10mo ago
Worth noting this is not only good on benchmarks, but significantly more efficient at inference https://x.com/_thomasip/status/1995489087386771851
29.
by red2awn 10mo ago
Check the maximum number of completion tokens allowed.
30.
by red2awn 10mo ago
Holy em-dash fest in the examples, would have thought they'd augment the training dataset to reduce this behavior.

More ›