INDEX

Explanations

phrases related to contradictions or conflicts

oai_token-act-pair · gpt-3.5-turbo

terms related to contradiction or conflict

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

NetMessage

-0.77

 throats

-0.72

istics

-0.70

eele

-0.69

 Mehran

-0.68

 GOODMAN

-0.65

 Nanto

-0.65

 Assass

-0.64

doms

-0.64

 Parenthood

-0.63

POSITIVE LOGITS

ptions

1.10

 contra

1.06

ption

1.05

ven

0.93

asca

0.88

ventions

0.83

vention

0.77

offensive

0.76

ctr

0.73

xon

0.73

Activations Density 0.015%

phrases related to contradictions or conflicts

terms related to contradiction or conflict

No Comments

No Known Activations