INDEX

Explanations

phrases related to legal proceedings or courts

oai_token-act-pair · gpt-3.5-turbo

references to the US Court of Appeals

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

xy

-0.69

 bare

-0.66

 save

-0.65

 crate

-0.65

gig

-0.63

 Lloyd

-0.62

jar

-0.62

ruff

-0.61

rade

-0.61

 stre

-0.61

POSITIVE LOGITS

 Appeals

3.82

 Appeal

2.19

 appeals

1.94

 appellate

1.53

 appeal

1.36

 appealed

1.27

 Claims

1.13

 Judges

1.09

 Judicial

1.07

eals

1.07

Activations Density 0.015%

phrases related to legal proceedings or courts

references to the US Court of Appeals

No Comments

No Known Activations