INDEX

Explanations

phrases related to comparisons or evaluations, particularly emphasizing a contrast between two elements

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "considering."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inis

-0.79

ernal

-0.78

uala

-0.77

scribe

-0.77

jer

-0.76

orem

-0.75

arez

-0.75

rouse

-0.73

vous

-0.73

inals

-0.71

POSITIVE LOGITS

how

0.93

why

0.72

 hindsight

0.70

 hordes

0.70

 recent

0.70

 everything

0.68

 what

0.66

 everyone

0.66

 that

0.65

 considering

0.65

Activations Density 0.082%

phrases related to comparisons or evaluations, particularly emphasizing a contrast between two elements

phrases that include the word "considering."

No Comments

No Known Activations