INDEX

Explanations

years from the 1970s

oai_token-act-pair · gpt-3.5-turbo

references to the year 1973 and its surrounding context

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

tro

-0.78

ndra

-0.74

Edge

-0.71

 entirety

-0.71

hop

-0.71

Storage

-0.67

unw

-0.66

edo

-0.65

BOOK

-0.64

unders

-0.63

POSITIVE LOGITS

ĸļ

0.74

 onwards

0.71

 onward

0.69

0.67

å¹

0.66

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

0.66

0.64

 Manson

0.64

sburg

0.64

 Jong

0.63

Activations Density 0.024%

years from the 1970s

references to the year 1973 and its surrounding context

No Comments

No Known Activations