INDEX

Explanations

email-related terms within text

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "email" and related terms

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

TRY

-0.79

ICLE

-0.78

-0.75

Jer

-0.74

liga

-0.68

GBT

-0.67

icle

-0.67

NEY

-0.66

ktop

-0.65

aft

-0.65

POSITIVE LOGITS

 inbox

1.16

 address

1.10

 addresses

1.01

 correspondence

0.92

email

0.92

address

0.91

 Address

0.85

 invitation

0.80

 newsletters

0.80

bag

0.79

Activations Density 0.029%

email-related terms within text

occurrences of the word "email" and related terms

No Comments

No Known Activations