INDEX

Explanations

terminology related to military attacks or negotiations

oai_token-act-pair · gpt-3.5-turbo

references to military or organized actions involving "strike."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gy

-0.73

 Categories

-0.67

ular

-0.66

 Goth

-0.64

Ħ

-0.62

icone

-0.61

 Addiction

-0.61

online

-0.61

 UCHIJ

-0.60

ohyd

-0.60

POSITIVE LOGITS

 strike

3.87

 strikes

2.76

strike

2.57

 Strike

2.56

Strike

2.19

 Strikes

1.99

 strikers

1.97

 struck

1.81

 striking

1.74

rikes

1.52

Activations Density 0.007%

terminology related to military attacks or negotiations

references to military or organized actions involving "strike."

No Comments

No Known Activations