INDEX

Explanations

references to beliefs or opinions

oai_token-act-pair · gpt-3.5-turbo

repeated use of the word "the"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ibaba

-0.78

ãĤ´ãĥ³

-0.75

imi

-0.73

bg

-0.73

icia

-0.71

strate

-0.70

Alert

-0.69

plete

-0.69

ftime

-0.69

ategory

-0.68

POSITIVE LOGITS

 slightest

1.20

 majority

1.09

 entire

1.05

 latter

0.98

 vast

0.97

 greatest

0.97

 absence

0.97

 whole

0.94

 biggest

0.92

 Russians

0.92

Activations Density 0.390%

references to beliefs or opinions

repeated use of the word "the"

No Comments

No Known Activations