INDEX

Explanations

IP addresses and network-related information

oai_token-act-pair · gpt-3.5-turbo

numerical data, particularly IP addresses and version information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 scrut

-0.70

ieth

-0.66

 downs

-0.65

lest

-0.64

viz

-0.63

emort

-0.62

 suspic

-0.62

aciously

-0.62

anchester

-0.61

mast

-0.61

POSITIVE LOGITS

 Fiat

0.77

 Prediction

0.67

 Simulation

0.66

#$

0.66

 RELEASE

0.65

 istg

0.65

 Trave

0.65

çļĦ

0.63

 Yesterday

0.63

 Collider

0.62

Activations Density 0.123%

IP addresses and network-related information

numerical data, particularly IP addresses and version information

No Comments

No Known Activations