INDEX

Explanations

numerical sequences such as phone numbers or license plate numbers

oai_token-act-pair · gpt-3.5-turbo

references to various types of identification numbers and related information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

reci

-0.78

ittle

-0.76

ipation

-0.75

duration

-0.72

issance

-0.72

ulp

-0.72

EMENT

-0.71

ISON

-0.70

igion

-0.69

ossom

-0.67

POSITIVE LOGITS

plates

0.90

 redacted

0.90

 spoof

0.84

 dial

0.84

 registry

0.83

 belonging

0.81

 Aadhaar

0.79

 scanner

0.78

 registration

0.77

XXXX

0.75

Activations Density 0.132%

numerical sequences such as phone numbers or license plate numbers

references to various types of identification numbers and related information

No Comments

No Known Activations