INDEX

Explanations

mentions of events or actions being described as called or calling

oai_token-act-pair · gpt-3.5-turbo

instances of the word "called" indicating references or mentions within the text

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oult

-0.91

IER

-0.83

ãĥ¯ãĥ³

-0.74

////////////////

-0.68

åŃĲ

-0.67

lean

-0.65

pmwiki

-0.63

vic

-0.62

restrial

-0.62

Place

-0.61

POSITIVE LOGITS

phas

0.82

 them

0.81

 himself

0.76

0.75

it

0.73

 attention

0.71

 aloud

0.70

him

0.69

terday

0.66

 homosexuality

0.66

Activations Density 0.055%

mentions of events or actions being described as called or calling

instances of the word "called" indicating references or mentions within the text

No Comments

No Known Activations