INDEX

Explanations

proper nouns related to people or places

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "work" in various forms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ylon

-0.81

 constitu

-0.70

idium

-0.70

 Ukrain

-0.67

emonic

-0.64

 Flavoring

-0.62

iren

-0.61

COMPLE

-0.60

fal

-0.60

 Shal

-0.59

POSITIVE LOGITS

 ethic

1.13

hops

1.04

flows

0.96

 collabor

0.94

 arrang

0.93

station

0.91

mates

0.89

bench

0.89

hirt

0.87

 tirelessly

0.87

Activations Density 0.079%

proper nouns related to people or places

repeated mentions of the word "work" in various forms

No Comments

No Known Activations