INDEX

Explanations

violent or conflict-related words, particularly related to arming and disarming

oai_token-act-pair · gpt-3.5-turbo

terms related to military or weaponry

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

order

-0.70

haps

-0.67

TBD

-0.67

 dime

-0.62

Hyper

-0.62

RTX

-0.61

DX

-0.60

ml

-0.60

sell

-0.59

 Skydragon

-0.59

POSITIVE LOGITS

arming

0.89

ageddon

0.84

ingly

0.79

xual

0.78

armed

0.78

ament

0.77

ishly

0.77

emort

0.76

eering

0.76

achine

0.75

Activations Density 0.011%

violent or conflict-related words, particularly related to arming and disarming

terms related to military or weaponry

No Comments

No Known Activations