INDEX

Explanations

names followed by adverbs modifying time

oai_token-act-pair · gpt-3.5-turbo

references to recent events or actions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ILCS

-0.70

breeding

-0.66

abiding

-0.61

 Adults

-0.61

ween

-0.60

intensity

-0.59

Textures

-0.59

aimon

-0.58

 preference

-0.58

Exper

-0.57

POSITIVE LOGITS

 tweeted

1.48

 testified

1.27

 retweet

1.21

 tweeting

1.19

 penned

1.18

 bluntly

1.17

 emailed

1.16

 angrily

1.15

 wrote

1.14

 commented

1.13

Activations Density 0.337%

names followed by adverbs modifying time

references to recent events or actions

No Comments

No Known Activations