INDEX

Explanations

the word "over" followed by a number indicating a level of intensity or duration

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "over."

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

uum

-0.70

oke

-0.67

ORY

-0.63

egal

-0.62

 Distribut

-0.60

ittal

-0.60

oker

-0.59

UE

-0.59

arium

-0.59

KE

-0.59

POSITIVE LOGITS

hang

1.19

tones

1.04

comes

1.04

drive

1.01

grown

0.99

haul

0.99

lord

0.98

clock

0.96

stay

0.93

blown

0.93

Activations Density 0.083%

the word "over" followed by a number indicating a level of intensity or duration

occurrences of the word "over."

No Comments

No Known Activations