INDEX

Explanations

phrases or terms related to central or core locations within various contexts

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a central or significant concept in a context

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

/-

-0.79

EStream

-0.78

BIL

-0.77

clips

-0.72

 Cosponsors

-0.71

Subscribe

-0.66

DragonMagazine

-0.66

PI

-0.65

 WATCHED

-0.64

é¾įå

-0.63

POSITIVE LOGITS

 Appalach

0.68

 Protestant

0.67

 contention

0.66

stown

0.62

 darkness

0.60

 Huck

0.60

 humanity

0.60

drive

0.59

arching

0.59

hearted

0.59

Activations Density 0.081%

phrases or terms related to central or core locations within various contexts

phrases indicating a central or significant concept in a context

No Comments

No Known Activations