INDEX

Explanations

names or terms related to specific persons or entities

oai_token-act-pair · gpt-3.5-turbo

occurrences of names or identifiers related to users in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

erest

-0.86

olulu

-0.85

ccording

-0.80

ured

-0.74

clud

-0.68

urers

-0.65

raising

-0.64

 diverse

-0.64

 Pradesh

-0.62

uring

-0.61

POSITIVE LOGITS

lein

1.06

bilt

0.95

jee

0.90

idge

0.90

gren

0.88

witz

0.86

 Brothers

0.86

meyer

0.84

iffe

0.84

wald

0.82

Activations Density 0.120%

names or terms related to specific persons or entities

occurrences of names or identifiers related to users in various contexts

No Comments

No Known Activations