INDEX

Explanations

names written in different languages

oai_token-act-pair · gpt-3.5-turbo

specific numerical values and their relevance in the context presented

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

Flash

-0.64

 prematurely

-0.60

 braces

-0.59

 shuffle

-0.59

 shockingly

-0.57

Bloom

-0.57

 sticking

-0.56

advertisement

-0.56

 backdrop

-0.56

IUM

-0.56

POSITIVE LOGITS

©¶æ¥µ

0.89

arent

0.86

Ã©

0.79

Ãł

0.79

£ı

0.78

Ã³

0.78

Ã¶r

0.78

Ã¼

0.78

asse

0.77

aren

0.77

Activations Density 0.118%

names written in different languages

specific numerical values and their relevance in the context presented

No Comments

No Known Activations