INDEX

Explanations

phrases related to mental processes, such as knowing, thinking, hoping, and believing

oai_token-act-pair · gpt-3.5-turbo

phrases expressing knowledge, belief, or awareness

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

 Difference

-0.70

pdf

-0.68

enum

-0.64

version

-0.62

PokÃ©

-0.62

ena

-0.62

auri

-0.62

asse

-0.60

issue

-0.60

haw

-0.60

POSITIVE LOGITS

lessly

0.82

 strangers

0.74

 thereby

0.73

 lest

0.71

 anticipation

0.69

 unsuspecting

0.68

efully

0.67

 fors

0.65

 nothing

0.65

 nobody

0.64

Activations Density 0.154%

phrases related to mental processes, such as knowing, thinking, hoping, and believing

phrases expressing knowledge, belief, or awareness

No Comments

No Known Activations