INDEX

Explanations

names or identifiers mentioned in text

oai_token-act-pair · gpt-3.5-turbo

instances of identification or naming of individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

hibition

-0.73

 guiActiveUnfocused

-0.70

ocratic

-0.67

course

-0.66

ocracy

-0.64

akings

-0.62

ocrats

-0.62

rouse

-0.61

 brim

-0.59

idious

-0.58

POSITIVE LOGITS

 pseudonym

0.88

as

0.85

 surn

0.83

by

0.83

onyms

0.81

 anonymously

0.80

 locally

0.77

onym

0.77

 only

0.75

 unnamed

0.75

Activations Density 0.080%

names or identifiers mentioned in text

instances of identification or naming of individuals

No Comments

No Known Activations