INDEX

Explanations

time-related phrases, specifically those indicating a sequence of events

oai_token-act-pair · gpt-3.5-turbo

instances of the word "after" and related temporal phrases

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

artisan

-0.81

ãĥĥãĥĪ

-0.80

quartered

-0.77

atories

-0.75

taboola

-0.73

offs

-0.71

Detailed

-0.70

æ©

-0.70

olitical

-0.67

requires

-0.67

POSITIVE LOGITS

 realizing

1.24

 hearing

1.24

 reading

1.19

 seeing

1.13

 watching

1.13

 discovering

1.12

 noticing

1.12

 awhile

1.11

 witnessing

1.05

 listening

1.03

Activations Density 0.170%

time-related phrases, specifically those indicating a sequence of events

instances of the word "after" and related temporal phrases

No Comments

No Known Activations