INDEX

Explanations

requests for interaction or feedback in a text

oai_token-act-pair · gpt-3.5-turbo

punctuation and expressions of engagement or prompting feedback

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 awakening

-0.83

 acute

-0.79

 rescuing

-0.77

 corrosion

-0.73

Ļ½

-0.72

 lifes

-0.72

 reversing

-0.67

 footing

-0.67

 immersion

-0.67

emaker

-0.66

POSITIVE LOGITS

<|endoftext|>

1.11

 Feedback

1.04

:)

0.97

 Otherwise

0.97

 EDIT

0.95

 Comments

0.92

 Suggest

0.92

 Thanks

0.92

 ðŁĳ

0.91

;)

0.91

Activations Density 0.145%

requests for interaction or feedback in a text

punctuation and expressions of engagement or prompting feedback

No Comments

No Known Activations