INDEX

Explanations

programming-related keywords or symbols

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

invokeLater

-0.34

DBNull

-0.32

<eos>

-0.31

 line

-0.30

 dedicated

-0.30

all

-0.30

me

-0.28

 Hold

-0.28

↵↵

-0.27

glo

-0.26

POSITIVE LOGITS

 Chwiliwch

0.84

 Normdatei

0.75

 nakalista

0.75

 kasarigan

0.74

Билгалдахарш

0.70

 CreateTagHelper

0.68

RTLR

0.65

 يتيمه

0.63

 ddelweddau

0.62

<unused8>

0.60

Activations Density 0.009%

programming-related keywords or symbols

No Known Activations