INDEX

Explanations

the word "impossible"

oai_token-act-pair · gpt-3.5-turbo

phrases that convey the concept of impossibility

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

avia

-0.94

erva

-0.80

rika

-0.78

uay

-0.76

lance

-0.75

elle

-0.74

aina

-0.74

Kings

-0.74

ript

-0.73

oola

-0.72

POSITIVE LOGITS

ossible

0.77

 feas

0.77

 unnecess

0.75

 simultane

0.73

 unanim

0.73

itability

0.71

ãĥ¬

0.71

istically

0.69

 destro

0.68

 tampering

0.68

Activations Density 0.023%

the word "impossible"

phrases that convey the concept of impossibility

No Comments

No Known Activations