INDEX

Explanations

phrases indicating the audience should continue following for further information or updates

oai_token-act-pair · gpt-3.5-turbo

phrases urging the audience to remain updated or informed

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

vernment

-0.72

isson

-0.69

NetMessage

-0.69

ocr

-0.64

 terminating

-0.64

 unilaterally

-0.63

 parting

-0.62

 expulsion

-0.61

 downfall

-0.61

 suppl

-0.60

POSITIVE LOGITS

 tuned

1.30

 classy

0.92

 alert

0.87

 informed

0.87

 Calm

0.87

 updated

0.86

 Updated

0.85

 Alert

0.85

Alert

0.85

 awake

0.84

Activations Density 0.023%

phrases indicating the audience should continue following for further information or updates

phrases urging the audience to remain updated or informed

No Comments

No Known Activations