INDEX

Explanations

phrases related to specific places or objects being revered

oai_token-act-pair · gpt-3.5-turbo

references to places, particularly memorials and landmarks

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

BILL

-0.75

arily

-0.75

paces

-0.72

fb

-0.71

binary

-0.69

maps

-0.67

iland

-0.67

issues

-0.66

blogs

-0.66

 restrooms

-0.66

POSITIVE LOGITS

 Excellence

0.96

 Sacrifice

0.92

Saint

0.88

 Allaah

0.87

 Minerva

0.84

 Darkness

0.84

 Wisdom

0.83

 Ascension

0.82

 Saint

0.82

Dam

0.80

Activations Density 0.123%

phrases related to specific places or objects being revered

references to places, particularly memorials and landmarks

No Comments

No Known Activations