INDEX

Explanations

names of individuals, locations, and events related to news articles and crime reports

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

instances of the letter "a" or variations of its presence in words

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

 whatsoever

-0.63

 unavoid

-0.63

eru

-0.63

 emerging

-0.61

 internationally

-0.60

ads

-0.60

 seminars

-0.59

 associations

-0.59

 events

-0.59

Ads

-0.58

POSITIVE LOGITS

 fortnight

1.31

decade

1.29

 month

1.26

 week

1.22

few

1.21

 year

1.19

 couple

1.16

day

1.06

flame

1.04

pron

1.01

Activations Density 0.101%

names of individuals, locations, and events related to news articles and crime reports

instances of the letter "a" or variations of its presence in words

No Comments

No Known Activations