INDEX

Explanations

words related to cutting or separating something into parts

oai_token-act-pair · gpt-3.5-turbo

repeated patterns or sequences in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

SHIP

-0.69

 prisoners

-0.64

 inmates

-0.61

 embodiments

-0.61

brate

-0.58

 labour

-0.58

 phyl

-0.57

 reconciliation

-0.57

theless

-0.57

 gulf

-0.56

POSITIVE LOGITS

ops

1.26

yright

1.01

heet

0.99

icle

0.91

oppers

0.91

imus

0.90

ilon

0.90

icles

0.89

hel

0.88

iate

0.88

Activations Density 0.011%

words related to cutting or separating something into parts

repeated patterns or sequences in the text

No Comments

No Known Activations