INDEX

Explanations

names of news agencies or individuals accompanied by specific symbols or initials

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 Squadron

-0.74

Dyn

-0.71

 Draco

-0.63

 Frozen

-0.63

 Penny

-0.61

 congr

-0.60

sbm

-0.60

cation

-0.60

 defaults

-0.59

 deposits

-0.58

POSITIVE LOGITS

ampa

0.96

ccording

0.94

SAN

0.91

WASHINGTON

0.88

Washington

0.84

ÂłÂłÂłÂłÂłÂłÂłÂł

0.84

Jer

0.82

SHARE

0.81

Madison

0.79

LOS

0.78

Activations Density 0.090%

names of news agencies or individuals accompanied by specific symbols or initials

No Comments

No Known Activations