INDEX

Explanations

the exact word "word" occurring in the text

oai_token-act-pair · gpt-3.5-turbo

multiple instances of the term "word" 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 @ 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

psey

-0.95

etheus

-0.74

opausal

-0.73

orsche

-0.73

âĹ¼

-0.70

ithing

-0.70

aples

-0.70

roxy

-0.69

ruciating

-0.69

ervation

-0.67

POSITIVE LOGITS

 word

1.03

sworth

0.97

press

0.97

naire

0.89

Word

0.88

 Word

0.87

word

0.75

lore

0.74

mith

0.73

 aloud

0.73

Activations Density 0.012%

the exact word "word" occurring in the text

multiple instances of the term "word" in various contexts

No Comments

No Known Activations