INDEX

Explanations

numerals followed by hyphens, usually indicating a range

oai_token-act-pair · gpt-3.5-turbo

quantifiable statistics or data presented in a numerical format

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 turbulence

-0.74

 Cassidy

-0.72

 scenery

-0.65

NV

-0.63

 Brav

-0.62

 Burr

-0.62

Uni

-0.60

Osw

-0.59

 transc

-0.58

 popcorn

-0.58

POSITIVE LOGITS

five

1.88

eight

1.84

seven

1.83

six

1.80

nine

1.79

four

1.72

three

1.71

two

1.64

Eight

1.37

fifth

1.34

Activations Density 0.023%

numerals followed by hyphens, usually indicating a range

quantifiable statistics or data presented in a numerical format

No Comments

No Known Activations