INDEX

Explanations

phrases related to opinions or judgments

oai_token-act-pair · gpt-3.5-turbo

expressions of personal opinions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

zz

-0.78

escape

-0.74

resp

-0.66

++;

-0.66

 Stretch

-0.62

zees

-0.62

ackle

-0.61

jar

-0.61

 Thief

-0.60

rises

-0.60

POSITIVE LOGITS

 opinion

3.89

 Opinion

2.63

 opin

2.55

 opinions

2.53

inion

1.68

 sentiment

1.46

 judgement

1.43

 judgment

1.37

 impression

1.31

 views

1.29

Activations Density 0.014%

phrases related to opinions or judgments

expressions of personal opinions

No Comments

No Known Activations