INDEX

Explanations

phrases or words related to close monitoring or association

oai_token-act-pair · gpt-3.5-turbo

references to being closely associated or connected

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Bucket

-0.79

 Lightning

-0.77

 Chaser

-0.76

ICAN

-0.71

Snow

-0.67

XL

-0.67

å£

-0.65

ane

-0.64

Ott

-0.64

ILLE

-0.63

POSITIVE LOGITS

 closely

1.10

 aligned

0.89

 resembles

0.89

 scrutin

0.86

 resembled

0.82

 enough

0.80

 akin

0.80

 traced

0.79

 monitored

0.78

 guarded

0.77

Activations Density 0.005%

phrases or words related to close monitoring or association

references to being closely associated or connected

No Comments

No Known Activations