INDEX

Explanations

phrases with social or political relevance

oai_token-act-pair · gpt-3.5-turbo

symbol sequences or non-standard characters

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

 Downs

-0.71

 Practices

-0.68

 straw

-0.66

 tremend

-0.64

 decomp

-0.63

ãĥīãĥ©ãĤ´ãĥ³

-0.63

 whistle

-0.62

 dispers

-0.62

 Pavilion

-0.62

 assemb

-0.61

POSITIVE LOGITS

į

1.03

ı

0.99

¤

0.98

ł

0.97

Ķ

0.96

«

0.91

¶

0.90

Ĥ

0.90

¬

0.90

±

0.90

Activations Density 0.117%

phrases with social or political relevance

symbol sequences or non-standard characters

No Comments

No Known Activations