INDEX

Explanations

references to oneself or reflexive actions

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "self."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

akings

-0.76

microsoft

-0.66

 airspace

-0.64

 Amend

-0.63

 wedge

-0.62

MSN

-0.58

 Learns

-0.58

osa

-0.57

 arrivals

-0.57

rought

-0.56

POSITIVE LOGITS

ortium

1.09

selves

1.04

destruct

0.92

self

0.84

ridges

0.83

theless

0.83

terday

0.81

same

0.75

ridge

0.73

acid

0.73

Activations Density 0.014%

references to oneself or reflexive actions

references to the concept of "self."

No Comments

No Known Activations