INDEX

Explanations

links or prompts directing to additional information or sources

oai_token-act-pair · gpt-3.5-turbo

instances of the word "here" as a call to action or to direct readers to additional content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Ļ½

-0.88

 manif

-0.81

 overtime

-0.80

onga

-0.79

 awakening

-0.78

 footing

-0.78

¶ħ

-0.76

untarily

-0.73

©¶æ¥µ

-0.72

 involuntary

-0.72

POSITIVE LOGITS

<|endoftext|>

1.05

aspx

1.02

 Thanks

1.02

 Also

1.00

 Though

0.96

 Subscribe

0.95

 Please

0.95

 Additionally

0.94

 Alternatively

0.92

 Retrieved

0.92

Activations Density 0.419%

links or prompts directing to additional information or sources

instances of the word "here" as a call to action or to direct readers to additional content

No Comments

No Known Activations