INDEX

Explanations

the text string 'rog' with varying degrees of emphasis

oai_token-act-pair · gpt-3.5-turbo

occurrences of the term "rogue"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 USPS

-0.67

 Speedway

-0.65

Â¢

-0.63

 Kindle

-0.62

-0.61

 Claus

-0.60

 pinch

-0.59

 birth

-0.58

ISO

-0.58

POSITIVE LOGITS

rog

1.64

raphic

1.07

aming

0.97

raphics

0.94

allery

0.94

raph

0.93

rogram

0.91

eln

0.89

hett

0.88

ratulations

0.88

Activations Density 0.005%

the text string 'rog' with varying degrees of emphasis

occurrences of the term "rogue"

No Comments

No Known Activations