INDEX

Explanations

mention of violent actions and incidents, specifically involving shooting and killing

oai_token-act-pair · gpt-3.5-turbo

phrases related to violent incidents and casualties

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

heit

-0.88

 Administ

-0.83

 Username

-0.79

ional

-0.75

SPONSORED

-0.72

soType

-0.72

VERTISEMENT

-0.66

TERN

-0.66

 Factor

-0.65

 Granted

-0.65

POSITIVE LOGITS

 torped

0.86

 espresso

0.85

 sparks

0.79

 lasers

0.76

 bullets

0.71

 bang

0.71

 Shotgun

0.70

 unarmed

0.70

 missiles

0.70

 cannon

0.69

Activations Density 0.148%

mention of violent actions and incidents, specifically involving shooting and killing

phrases related to violent incidents and casualties

No Comments

No Known Activations