INDEX

Explanations

details related to a crime scene, involving police activity and violent actions

oai_token-act-pair · gpt-3.5-turbo

vocabulary related to crime and violence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Trade

-0.75

ãĤ¨

-0.70

omorphic

-0.69

Economic

-0.64

ophon

-0.63

astical

-0.63

ãĤ»

-0.62

Craft

-0.62

ptions

-0.62

onomy

-0.61

POSITIVE LOGITS

 brig

0.80

 altercation

0.80

 sheriff

0.78

 taped

0.77

 NYPD

0.75

 handcuffed

0.75

 distraught

0.74

 saddened

0.74

 arra

0.74

 raided

0.74

Activations Density 3.329%

details related to a crime scene, involving police activity and violent actions

vocabulary related to crime and violence

No Comments

No Known Activations