INDEX

Explanations

sequences of numbers with different format characters in between

oai_token-act-pair · gpt-3.5-turbo

numerical values or references to statistics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 teasp

-0.74

vertisement

-0.69

BuyableInstoreAndOnline

-0.67

 corrid

-0.66

 comprom

-0.65

PDATE

-0.63

 Leban

-0.62

 referen

-0.61

 carbohyd

-0.61

gdala

-0.59

POSITIVE LOGITS

wm

1.01

df

0.86

0.85

chev

0.83

ffe

0.81

eus

0.78

eff

0.76

bg

0.75

rez

0.74

Activations Density 0.102%

sequences of numbers with different format characters in between

numerical values or references to statistics

No Comments

No Known Activations