INDEX

Explanations

words related to retreating or treating, often in medical or military contexts

oai_token-act-pair · gpt-3.5-turbo

terms related to recreation or enjoyable activities

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

OTA

-0.69

BO

-0.69

 Wilde

-0.66

BIL

-0.63

NDP

-0.62

BOX

-0.61

 siph

-0.61

 Scotia

-0.60

PLIC

-0.60

 Moto

-0.60

POSITIVE LOGITS

htaking

1.03

terness

1.01

reat

0.99

idon

0.97

sburg

0.97

resses

0.96

ures

0.94

hens

0.92

iful

0.89

ottesville

0.89

Activations Density 0.009%

words related to retreating or treating, often in medical or military contexts

terms related to recreation or enjoyable activities

No Comments

No Known Activations