INDEX

Explanations

mentions of the specific term "Wire"

oai_token-act-pair · gpt-3.5-turbo

references to a specific media outlet or publication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 replicate

-0.65

moderate

-0.64

nas

-0.64

 administer

-0.63

interested

-0.62

 replication

-0.62

 superst

-0.62

 moderate

-0.62

 carniv

-0.62

 cause

-0.62

POSITIVE LOGITS

 Wire

4.37

Wire

2.72

wire

1.85

 wire

1.53

 Wireless

1.24

 wires

1.20

 Cable

1.19

 wiret

1.07

 Tape

1.05

 Router

1.05

Activations Density 0.021%

mentions of the specific term "Wire"

references to a specific media outlet or publication

No Comments

No Known Activations