INDEX

Explanations

the concept of an "end" or a conclusion in a narrative

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate conclusions or endpoints

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

æ©Ł

-0.72

thouse

-0.68

BLIC

-0.64

hee

-0.63

itsch

-0.63

Dou

-0.61

 Tanks

-0.61

PDATE

-0.60

KER

-0.60

£ı

-0.59

POSITIVE LOGITS

angering

1.34

owment

1.27

angers

1.03

angered

1.03

ocrin

0.96

orph

0.96

ocrine

0.96

game

0.90

urance

0.87

eared

0.85

Activations Density 0.046%

the concept of an "end" or a conclusion in a narrative

phrases that indicate conclusions or endpoints

No Comments

No Known Activations