INDEX

Explanations

phrases related to categorization or comparison

oai_token-act-pair · gpt-3.5-turbo

phrases indicating membership or inclusion within a group or concept

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

ahime

-0.83

lyak

-0.74

 Locations

-0.61

iland

-0.61

berry

-0.61

ischer

-0.61

ortment

-0.60

ugi

-0.60

BAT

-0.60

soType

-0.59

POSITIVE LOGITS

 whatsoever

1.49

nor

1.33

 anymore

1.00

 dime

0.95

 except

0.91

 necessarily

0.85

 bothered

0.83

 slightest

0.83

 EVER

0.80

 anybody

0.79

Activations Density 0.131%

phrases related to categorization or comparison

phrases indicating membership or inclusion within a group or concept

No Comments

No Known Activations