INDEX

Explanations

references to locations or positions, especially on opposite sides or ends

oai_token-act-pair · gpt-3.5-turbo

instances of the word "end" and its variations in different contexts

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

issance

-0.75

PDATE

-0.75

 Flavoring

-0.74

dinand

-0.72

atoon

-0.63

ndum

-0.62

éĹĺ

-0.62

æ©Ł

-0.61

 irregular

-0.60

veyard

-0.59

POSITIVE LOGITS

angering

1.29

owment

1.19

angers

1.01

ocrine

0.98

ocrin

0.95

anger

0.87

game

0.86

orph

0.83

points

0.83

tails

0.82

Activations Density 0.041%

references to locations or positions, especially on opposite sides or ends

instances of the word "end" and its variations in different contexts

No Comments

No Known Activations