INDEX

Explanations

strings of characters like ">>"

oai_token-act-pair · gpt-3.5-turbo

symbols and formatting related to digital or online content

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

 dispers

-0.74

 Vaugh

-0.73

 Bengal

-0.69

 Berm

-0.68

hetti

-0.67

ured

-0.67

laus

-0.66

quel

-0.65

rish

-0.64

edo

-0.64

POSITIVE LOGITS

SOURCE

0.91

[[

0.90

MORE

0.89

_>

0.88

<<

0.81

wcsstore

0.79

QUEST

0.78

Â¢

0.76

>>

0.75

nery

0.74

Activations Density 0.013%

strings of characters like ">>"

symbols and formatting related to digital or online content

No Comments

No Known Activations