INDEX

Explanations

words that contain the substring 'te' with varying activation strengths indicating a preference for longer or more specific matches

oai_token-act-pair · gpt-3.5-turbo

references to "teetering" or related concepts of instability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 demographics

-0.67

 Timberwolves

-0.67

 Hawks

-0.62

stood

-0.60

OV

-0.59

 Nationals

-0.59

INGTON

-0.59

 Predators

-0.59

 tolerated

-0.58

 Magnus

-0.58

POSITIVE LOGITS

legraph

1.35

legram

1.28

pee

1.19

cn

1.18

eming

1.11

asers

1.06

quila

1.03

pees

1.01

achable

0.93

chie

0.89

Activations Density 0.020%

words that contain the substring 'te' with varying activation strengths indicating a preference for longer or more specific matches

references to "teetering" or related concepts of instability

No Comments

No Known Activations