INDEX

Explanations

names of political figures and government-related terms

oai_token-act-pair · gpt-3.5-turbo

references to government entities and legalities

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

 Horus

-0.76

 Whilst

-0.73

 Highlander

-0.73

 Tosh

-0.72

 Jagu

-0.72

 Alic

-0.69

 Trance

-0.69

 surn

-0.68

 Celt

-0.68

 Belg

-0.67

POSITIVE LOGITS

 memos

0.88

amacare

0.85

igroup

0.83

asury

0.83

 memo

0.83

HHS

0.81

FDA

0.77

ewater

0.76

 attorney

0.76

 lawsuits

0.76

Activations Density 0.225%

names of political figures and government-related terms

references to government entities and legalities

No Comments

No Known Activations