INDEX

Explanations

statements expressing strong opinions or evaluations, often with emphasis

oai_token-act-pair · gpt-3.5-turbo

assertions about political situations or narratives, particularly around accountability and perspectives on actions taken

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

 tactile

-0.87

 flavours

-0.78

ktop

-0.74

 handset

-0.71

 pastoral

-0.69

 experimental

-0.69

 licences

-0.69

eger

-0.69

 occasion

-0.68

 explor

-0.68

POSITIVE LOGITS

Hillary

1.29

Republicans

1.24

Democrats

1.23

 democr

1.22

TRUMP

1.22

GOP

1.21

Bernie

1.19

 Dems

1.17

 AMERICA

1.16

Trump

1.13

Activations Density 0.799%

statements expressing strong opinions or evaluations, often with emphasis

assertions about political situations or narratives, particularly around accountability and perspectives on actions taken

No Comments

No Known Activations