INDEX

Explanations

phrases related to events or announcements marked with "First."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "First"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

================================================================

-0.70

vg

-0.68

ribly

-0.65

 absor

-0.63

ucl

-0.62

$.

-0.62

 otherwise

-0.62

phant

-0.61

rolet

-0.61

rarily

-0.58

POSITIVE LOGITS

 Published

1.06

 responders

0.88

 Appearance

0.86

 impressions

0.86

 Nations

0.85

 Steps

0.82

 published

0.82

 baseman

0.80

 Lady

0.79

 lady

0.76

Activations Density 0.038%

phrases related to events or announcements marked with "First."

instances of the word "First"

No Comments

No Known Activations