INDEX

Explanations

proper nouns

oai_token-act-pair · gpt-3.5-turbo

keywords related to specific individuals and their influence or actions

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

 logical

-0.70

×©

-0.70

é¾

-0.68

 Bulg

-0.65

 Grid

-0.63

 metabolism

-0.62

worthiness

-0.60

WARD

-0.60

 Decay

-0.60

çĦ

-0.59

POSITIVE LOGITS

quist

1.60

gren

1.02

acion

0.90

ensen

0.89

thal

0.89

 submar

0.85

inav

0.84

bilt

0.82

ophon

0.82

qv

0.82

Activations Density 0.006%

proper nouns

keywords related to specific individuals and their influence or actions

No Comments

No Known Activations