INDEX

Explanations

phrases or statements indicating truth or accuracy

oai_token-act-pair · gpt-3.5-turbo

assertions or statements about truthfulness or conditions being true or false

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ngth

-0.81

fect

-0.78

afia

-0.77

udeb

-0.72

 hail

-0.69

umbnails

-0.67

ividual

-0.67

cies

-0.65

izabeth

-0.65

elve

-0.64

POSITIVE LOGITS

 happening

1.20

true

0.90

 true

0.88

 happen

0.88

 occurring

0.87

 untrue

0.86

 achievable

0.84

 TRUE

0.82

 accomplished

0.82

 happ

0.81

Activations Density 0.249%

phrases or statements indicating truth or accuracy

assertions or statements about truthfulness or conditions being true or false

No Comments

No Known Activations