INDEX

Explanations

phrases related to significant historical events or incidents

oai_token-act-pair · gpt-3.5-turbo

references to attacks and violent actions against individuals or entities

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

igmat

-0.76

 Canaver

-0.69

iasis

-0.69

reed

-0.68

 caveats

-0.67

idth

-0.67

ozo

-0.65

 curv

-0.65

awaru

-0.64

rang

-0.64

POSITIVE LOGITS

 targets

0.88

 convoy

0.84

 innocent

0.83

 premises

0.82

 unarmed

0.77

 belonging

0.75

 frequ

0.75

 guarding

0.75

 innoc

0.74

 buildings

0.72

Activations Density 0.386%

phrases related to significant historical events or incidents

references to attacks and violent actions against individuals or entities

No Comments

No Known Activations