INDEX

Explanations

names and code separators

np_acts-logits-general · gemini-2.5-flash-lite

structural and formatting cues typical of technical or scripted text—punctuation, symbols, code/path syntax, and generic glue words—rather than specific content terms.

oai_token-act-pair · gpt-5 Triggered by @vetterc0

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.3-70B-IT @ 50-resid-post-gf

Configuration

Goodfire/Llama-3.3-70B-Instruct-SAE-l50/Llama-3.3-70B-Instruct-SAE-l50.pt

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

%[

-0.08

 )\n\n\n\n\n\n\n\n

-0.08

Âł

-0.08

 _____

-0.08

âĢ¦

-0.07

ï½ŀï½ŀ

-0.07

's

-0.07

"(

-0.07

 [âĢ¦

-0.07

POSITIVE LOGITS

¶Į

0.12

ÂĢÂĢ

0.10

EMPLARY

0.10

¦æĥħ

0.10

******č\n

0.10

 -*-č\n

0.10

.RELATED

0.09

įng

0.09

 tieten

0.09

.showMessageDialog

0.09

Activations Density 3.356%

names and code separators

structural and formatting cues typical of technical or scripted text—punctuation, symbols, code/path syntax, and generic glue words—rather than specific content terms.

No Comments

No Known Activations

names and code separators

structural and formatting cues typical of technical or scripted text—punctuation, symbols, code/path syntax, and generic glue words—rather than specific content terms.

No Comments

No Known Activations