INDEX

Explanations

numerical entities like years and links in a text

oai_token-act-pair · gpt-3.5-turbo

references to notable dates and media sources

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

aint

-0.76

aries

-0.67

sty

-0.63

 antit

-0.59

hetti

-0.58

 brist

-0.58

 somet

-0.57

 undert

-0.57

Urs

-0.56

POSITIVE LOGITS

1.12

0.94

0.90

 Seventh

0.87

0.84

0.83

0.81

0.80

Activations Density 0.063%

numerical entities like years and links in a text

references to notable dates and media sources

No Comments

No Known Activations