INDEX

Explanations

references to the word "fox" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to foxes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Effective

-0.88

Rated

-0.76

itutional

-0.71

ETH

-0.71

affer

-0.67

Interstitial

-0.66

NT

-0.66

igious

-0.66

apter

-0.63

FINE

-0.63

POSITIVE LOGITS

es

1.02

fox

0.86

 squirrel

0.86

bat

0.82

manship

0.82

esy

0.81

hound

0.81

sey

0.76

fox

0.76

bench

0.75

Activations Density 0.018%

references to the word "fox" in various contexts

references to foxes

No Comments

No Known Activations