INDEX

Explanations

locations or organizations mentioned in news articles

oai_token-act-pair · gpt-3.5-turbo

instances of commas and punctuation in lists

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 shades

-0.65

 macros

-0.64

 exams

-0.63

terms

-0.62

 accounts

-0.62

 pointers

-0.61

tones

-0.61

 levels

-0.61

 cues

-0.61

 spores

-0.60

POSITIVE LOGITS

MEN

1.03

 CITY

0.96

 COURT

0.89

 DISTRICT

0.88

JUL

0.87

INC

0.84

INC

0.82

 FANTASY

0.82

 CHRIST

0.82

MEN

0.81

Activations Density 0.066%

locations or organizations mentioned in news articles

instances of commas and punctuation in lists

No Comments

No Known Activations