INDEX

Explanations

words used to identify people giving information in reports

oai_token-act-pair · gemini-2.0-flash

reporting verbs

np_token-act-pair-logits · gpt-4o-mini

source

np_max-act-logits · gemini-2.0-flash

New Auto-Interp

Configuration

google/gemma-scope-2b-pt-transcoders/layer_21/width_16k/average_l0_13

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.21.ln2.hook_normalized

Architecture

jumprelu_transcoder

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Theſe

-1.05

 Monfieur

-0.93

Efq

-0.86

 Beſ

-0.84

 Anſ

-0.84

 perfons

-0.83

 Houſe

-0.80

 snippetHide

-0.80

 Jefus

-0.79

 ſeveral

-0.76

POSITIVE LOGITS

 said

0.65

say

0.59

QtCore

0.53

 تق

0.50

0.49

qu

0.49

 stated

0.47

in

0.45

 Comp

0.45

 comp

0.45

Activations Density 1.463%

words used to identify people giving information in reports

reporting verbs

source

No Comments

No Known Activations