INDEX

Explanations

mentions related to text, possibly in the context of online content or programming

oai_token-act-pair · gpt-3.5-turbo

instances of the text command or placeholder within the document

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

IFrame

Link

Negative Logits

 Polaris

-0.70

 retired

-0.67

 ranks

-0.67

 Feder

-0.66

 rookie

-0.65

 retirement

-0.65

 surges

-0.63

 Carbuncle

-0.63

 stocks

-0.62

 rookies

-0.62

POSITIVE LOGITS

text

4.10

Text

2.28

 Text

2.07

TEXT

2.06

 text

2.05

 TEXT

1.84

 texts

1.61

 textual

1.39

texture

1.26

message

1.25

Activations Density 0.013%

mentions related to text, possibly in the context of online content or programming

instances of the text command or placeholder within the document

No Known Activations