INDEX

Explanations

phrases related to the action of shutting down

oai_token-act-pair · gpt-3.5-turbo

instances of shutdown-related terminology

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

xtap

-0.75

uld

-0.72

Pic

-0.69

erity

-0.66

é¾įå

-0.66

inen

-0.66

nder

-0.62

 amen

-0.61

MAT

-0.61

 Drawn

-0.61

POSITIVE LOGITS

river

0.74

 valves

0.70

grading

0.69

Interstitial

0.67

 freeway

0.67

 stairs

0.66

escent

0.65

 shop

0.64

 Firefly

0.63

ãĤ¬

0.63

Activations Density 0.022%

phrases related to the action of shutting down

instances of shutdown-related terminology

No Comments

No Known Activations