INDEX

Explanations

references to rankings or positions in a list

oai_token-act-pair · gpt-3.5-turbo

references to various rankings and their associated values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

vous

-0.94

gm

-0.81

cise

-0.76

romy

-0.74

Ã±a

-0.73

rote

-0.71

resp

-0.70

veh

-0.70

Ple

-0.68

pron

-0.67

POSITIVE LOGITS

 rankings

1.44

 Rankings

1.21

 ranking

1.06

 standings

0.94

Rank

0.92

uggest

0.87

elist

0.84

 ratings

0.83

 contenders

0.83

 listings

0.83

Activations Density 0.007%

references to rankings or positions in a list

references to various rankings and their associated values

No Comments

No Known Activations