INDEX

Explanations

references to the character "Frodo" or variations of the name

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names of characters and people

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

×ŀ

-0.83

×

-0.78

dress

-0.75

×Ļ

-0.72

loo

-0.72

ãĤ¨ãĥ«

-0.70

ãĤ´ãĥ³

-0.70

************

-0.70

creen

-0.68

****************

-0.68

POSITIVE LOGITS

unin

0.98

ansson

0.87

atche

0.86

adh

0.83

rites

0.82

auld

0.80

ringe

0.77

aldi

0.74

licks

0.74

adows

0.72

Activations Density 0.018%

references to the character "Frodo" or variations of the name

proper nouns, particularly names of characters and people

No Comments

No Known Activations