INDEX

Explanations

words related to the usage of items or products, specifically the phrase "used to"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "used" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

awk

-0.71

Hai

-0.67

nesty

-0.66

brother

-0.66

Ń·

-0.65

iversary

-0.63

COMPLE

-0.62

antle

-0.62

joice

-0.61

bow

-0.60

POSITIVE LOGITS

 Used

0.88

fully

0.80

FUL

0.79

 condoms

0.79

 sparing

0.78

 phrases

0.76

fulness

0.74

Used

0.71

nesday

0.71

 used

0.68

Activations Density 0.065%

words related to the usage of items or products, specifically the phrase "used to"

instances of the word "used" in various contexts

No Comments

No Known Activations