INDEX

Explanations

words related to specific entities or names, particularly ending in "-etta" with a strong preference for "vendetta."

oai_token-act-pair · gpt-3.5-turbo

mentions of specific names or entities, particularly related to a scientific context

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

Not in Any Lists

No Comments

Negative Logits

 private

-0.74

 exchanges

-0.67

 Crown

-0.66

 counts

-0.64

 places

-0.63

 fell

-0.63

 draw

-0.62

 strip

-0.62

 conspicuous

-0.61

 fall

-0.61

POSITIVE LOGITS

etta

5.15

ette

1.44

ett

1.37

eta

1.36

etti

1.34

ello

1.17

amera

1.15

elli

1.14

iola

1.09

aukee

1.08

Activations Density 0.019%

words related to specific entities or names, particularly ending in "-etta" with a strong preference for "vendetta."

mentions of specific names or entities, particularly related to a scientific context

No Comments

No Known Activations