INDEX

Explanations

mentions of a specific publication named "Observer"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Observer."

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

iling

-0.71

fil

-0.65

rid

-0.64

cial

-0.62

 slash

-0.62

ilings

-0.62

¾

-0.61

 imported

-0.59

 stretch

-0.59

vl

-0.58

POSITIVE LOGITS

 Observer

4.43

 observer

1.67

Obs

1.57

 Guardian

1.27

 Telegraph

1.19

Obs

1.15

 observers

1.11

 Economist

1.09

 Observatory

1.08

 Intercept

1.06

Activations Density 0.026%

mentions of a specific publication named "Observer"

instances of the word "Observer."

No Comments

No Known Activations