> did you read
Yeah its my blog. From a year ago. Your claim was “new capability” of frontier llms to have gained weight space understanding.
In the blog you would see the “edits”, the larger llm also reasons about target model weights. Here is an ex of Gemini 2.0 flash reasoning and coming up with weight edits for target qwen 0.6B to improve its HumanEval bench. Specifically its MLP block.
> different experiment
Any experiment that makes them reason through weights to come up with edits suggests understanding weight space existed long before.
1
28
Turn Qwen TTS perpetually angry
417 earlier mapped neurons, d_L ∈ ℝ¹⁰²⁴, L20–27
Add : v̂_L, L20–26, v̂ ⟂ d_L, v̂ ⟂ <eos>
h_L += 0.03·‖h_L‖·v̂_L every frame @ 12.5 Hz
read_direction: L8#1968↓ L10#888↓ L10#2660↑ L15#3057↓
Result :
f0 : 197 to 379 Hz
Spectral centroid shift: +1135
The Lazarus research by @DimitrisPapail should open fun interpretability objectives.
Neuron space editing Qwen3-TTS
Target : talker MLP neurons
Second-voice trait, β3 (400 mapped + add to activation before down_proj):
L23#1430 ↑ +5.287
L24#527 ↑ +1.674
L21#175 ↑ +0.315
L22#142 ↓ −2.010
L22#555 ↓ −0.220
Reprogramming dormant neurons : Qwen TTS
Ever wondered why Qwen3-TTS sometimes breathes or laughs out of nowhere? Agent found neurons behind it.
The details:
Agent found 417 neurons to be specific (about 0.48% of the talker's MLP neurons) and 61 of them are dormant. Amplified together, they give the following effects effect:
1. Breathes as words are spoken
2. Hushed ASMR
3. Natural laughs as language is being spoken
4. Enhanced whispering
5. 2nd speaker appearing.
Comparisons and neuron map repo in the comments below.
Inspired by recent work on Lazarus by @DimitrisPapail , which reprogrammed dormant neurons.
Invoking Whispering words at will with Neuron Space Edits in Qwen TTS.
Reprogramming dormant neurons : Qwen TTS
Ever wondered why Qwen3-TTS sometimes breathes or laughs out of nowhere? Agent found neurons behind it.
The details:
Agent found 417 neurons to be specific (about 0.48% of the talker's MLP neurons) and 61 of them are dormant. Amplified together, they give the following effects effect:
1. Breathes as words are spoken
2. Hushed ASMR
3. Natural laughs as language is being spoken
4. Enhanced whispering
5. 2nd speaker appearing.
Comparisons and neuron map repo in the comments below.
Inspired by recent work on Lazarus by @DimitrisPapail , which reprogrammed dormant neurons.
Given the Lazarus research, its obvious we need better tooling for observability and interpretability. I urge everyone to develop your own harnesses. One such harness in action of agent mapping speech pattern nuerons of Qwen TTS.
SOUND ON : If you are a Tolkien nerd.
Reprogramming dormant neurons : Qwen TTS
Ever wondered why Qwen3-TTS sometimes breathes or laughs out of nowhere? Agent found neurons behind it.
The details:
Agent found 417 neurons to be specific (about 0.48% of the talker's MLP neurons) and 61 of them are dormant. Amplified together, they give the following effects effect:
1. Breathes as words are spoken
2. Hushed ASMR
3. Natural laughs as language is being spoken
4. Enhanced whispering
5. 2nd speaker appearing.
Comparisons and neuron map repo in the comments below.
Inspired by recent work on Lazarus by @DimitrisPapail , which reprogrammed dormant neurons.
Someone else joins in. On a podcast-style voice: a "yeah", a reply, a question back. The dual speaking is NOT used in the reference audio yet it , upon neuron edits, chose to add those.
1
1
449
Laughing while talking. The laugh lands inside the words, not as a sound effect between them.
1
2
571



