INDEX

Explanations

phrases indicating a previous mention or reference in a text

oai_token-act-pair · gpt-3.5-turbo

references to prior studies or previous information

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

lua

-0.75

alker

-0.72

atron

-0.72

asp

-0.71

aliation

-0.70

rage

-0.69

istan

-0.69

ifle

-0.68

ocracy

-0.68

aleigh

-0.66

POSITIVE LOGITS

 generations

1.11

 incarn

1.06

 iterations

0.91

 incarnation

0.84

 administrations

0.84

 installments

0.82

ebin

0.82

 versions

0.80

 affili

0.79

 editions

0.79

Activations Density 0.027%

phrases indicating a previous mention or reference in a text

references to prior studies or previous information

No Comments

No Known Activations