INDEX

Explanations

sequences or items that have been ordered or ranked in a competitive or tournament-like context

oai_token-act-pair · gpt-3.5-turbo

sequences of repeated characters

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

uers

-0.82

ued

-0.80

ues

-0.72

Peb

-0.70

Js

-0.69

alg

-0.66

orse

-0.66

ilitating

-0.65

gers

-0.65

erness

-0.64

POSITIVE LOGITS

————

0.81

ãĤ¦ãĤ¹

0.73

gow

0.71

gomery

0.70

——

0.69

ially

0.69

DOWN

0.67

 Keynes

0.65

LEY

0.63

ellectual

0.63

Activations Density 0.044%

sequences or items that have been ordered or ranked in a competitive or tournament-like context

sequences of repeated characters

No Comments

No Known Activations