INDEX

Explanations

plural, often abstract, nouns—especially when appearing in lists, headings, or clause-heavy descriptions (e.g., with “that/which” constructions).

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

that

np_max-act · gpt-5 Triggered by @smilgaveronika

are

np_max-act-logits · gpt-5 Triggered by @smilgaveronika

New Auto-Interp

Configuration

google/gemma-scope-2-1b-pt/resid_post/layer_17_width_16k_l0_medium

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ತನ್ನ

1.62

was

1.57

 понял

1.48

 बचाता

1.41

Was

1.38

 தனது

1.38

Was

1.35

 தெரிவித்துள்ளது

1.34

 रोकता

1.33

 ஏற்படுத்துகிறது

1.33

POSITIVE LOGITS

are

3.89

 đều

3.82

都很

3.47

 were

3.42

都是

3.34

都有

3.24

 aren

3.21

 هستند

3.17

都在

3.17

 jsou

3.11

Activations Density 8.303%

plural, often abstract, nouns—especially when appearing in lists, headings, or clause-heavy descriptions (e.g., with “that/which” constructions).

that

are

No Comments

No Known Activations

plural, often abstract, nouns—especially when appearing in lists, headings, or clause-heavy descriptions (e.g., with “that/which” constructions).

that

are

No Comments

No Known Activations