INDEX

Explanations

explicit words or phrases

oai_token-act-pair · gpt-3.5-turbo

references to explicit content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.76

 Quincy

-0.76

pered

-0.71

Squ

-0.67

rug

-0.66

 Brow

-0.65

Spr

-0.65

 Tycoon

-0.65

kee

-0.64

ALD

-0.64

POSITIVE LOGITS

 explicit

1.16

 guiActiveUn

1.07

 explor

0.82

 textual

0.80

 explicitly

0.80

 nudity

0.75

 transmission

0.74

 implicit

0.74

foundland

0.73

 transmissions

0.72

Activations Density 0.007%

explicit words or phrases

references to explicit content

No Comments

No Known Activations