INDEX

Explanations

loading sequences, often found in web or video content

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Loading."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

stood

-0.78

ially

-0.76

Ĥİ

-0.73

atively

-0.73

hips

-0.70

beit

-0.67

fman

-0.67

pher

-0.65

Ĭ±

-0.64

liest

-0.63

POSITIVE LOGITS

 Loading

0.99

 Bagg

0.77

 Thumbnails

0.76

 guiName

0.74

 guiActive

0.72

 Load

0.72

 Rooms

0.71

 guiActiveUnfocused

0.70

Screen

0.70

 Tickets

0.69

Activations Density 0.012%

loading sequences, often found in web or video content

instances of the word "Loading."

No Comments

No Known Activations