INDEX

Explanations

words related to law enforcement incidents and legal actions

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

occurrences of the word "and" related to incidents involving violence or crime

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

isSpecialOrderable

-1.09

 Integration

-0.94

idth

-0.86

AES

-0.79

ablishment

-0.76

 Improvements

-0.76

oire

-0.74

parency

-0.73

iasm

-0.71

ngth

-0.70

POSITIVE LOGITS

 killed

1.38

 stabbed

1.26

 robbed

1.24

 wounded

1.13

 murdered

1.08

 paralyzed

1.08

 hospitalized

1.06

 choked

1.04

 died

1.04

 blinded

1.04

Activations Density 0.086%

words related to law enforcement incidents and legal actions

occurrences of the word "and" related to incidents involving violence or crime

No Comments

No Known Activations