INDEX

Explanations

phrases or sentences that are used to verify that a human is interacting with a system and is not a robot

oai_token-act-pair · gpt-3.5-turbo

references to robots or robotic concepts

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

ources

-0.67

 Oaks

-0.66

ciples

-0.64

 Tong

-0.64

 Emin

-0.63

drops

-0.63

 avenues

-0.62

ciation

-0.62

monds

-0.61

 Hudson

-0.61

POSITIVE LOGITS

ichick

0.76

 anymore

0.75

obic

0.73

 reliant

0.71

hered

0.69

ecast

0.68

proof

0.68

dayName

0.68

iframe

0.68

escent

0.66

Activations Density 0.020%

phrases or sentences that are used to verify that a human is interacting with a system and is not a robot

references to robots or robotic concepts

No Comments

No Known Activations