INDEX

Explanations

based on, presidential election

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

structural and formatting cues typical of technical or instructional text, including markup/metadata tokens, list/numbering and whitespace punctuation, and other indicators of formal document structure.

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

[,]

-0.09

 suce

-0.08

ï¼ıï¼ıï¼ıï¼ıï¼ıï¼ıï¼ıï¼ı

-0.07

à¸Ļà¸Ń

-0.07

[=

-0.07

rst

-0.06

 disappe

-0.06

dee

-0.06

\n

-0.06

POSITIVE LOGITS

¶Į

0.13

¦æĥħ

0.12

ÐĲÑĢÑħÑĸÐ²

0.11

----</

0.10

---</

0.10

printStats

0.09

ıa

0.09

ĥ½

0.09

'gc

0.09

ķãĤĵ

0.09

Activations Density 0.472%

based on, presidential election

structural and formatting cues typical of technical or instructional text, including markup/metadata tokens, list/numbering and whitespace punctuation, and other indicators of formal document structure.

No Comments

No Known Activations

based on, presidential election

structural and formatting cues typical of technical or instructional text, including markup/metadata tokens, list/numbering and whitespace punctuation, and other indicators of formal document structure.

No Comments

No Known Activations