INDEX

Explanations

phrases related to support or endorsement

oai_token-act-pair · gpt-3.5-turbo

instances of the word "back" or its variations related to support or defense

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

thora

-0.74

itizen

-0.69

entric

-0.69

nesota

-0.61

lys

-0.61

orp

-0.61

ifix

-0.60

pox

-0.60

è¦ļéĨĴ

-0.59

cz

-0.59

POSITIVE LOGITS

track

1.25

 away

0.95

ped

0.94

tracking

0.79

tr

0.77

INTON

0.77

up

0.77

dash

0.77

drive

0.74

 down

0.73

Activations Density 0.030%

phrases related to support or endorsement

instances of the word "back" or its variations related to support or defense

No Comments

No Known Activations