INDEX

Explanations

words related to providing feedback or criticism with the intention of improvement

oai_token-act-pair · gpt-3.5-turbo

terms related to constructive feedback and wilderness conservation

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

plex

-0.76

ighty

-0.69

gdala

-0.67

achy

-0.66

eph

-0.65

aeda

-0.65

bus

-0.63

clipse

-0.63

igans

-0.63

gary

-0.63

POSITIVE LOGITS

=]

0.88

ÃĥÃĤ

0.79

^^^^

0.77

 Liberties

0.76

ß

0.74

istered

0.73

////////

0.72

IZE

0.68

estern

0.66

zes

0.66

Activations Density 0.024%

words related to providing feedback or criticism with the intention of improvement

terms related to constructive feedback and wilderness conservation

No Comments

No Known Activations