INDEX

Explanations

pronouns and possessive adjectives

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

sson

-0.77

lund

-0.76

ocument

-0.74

urable

-0.72

spir

-0.67

erman

-0.66

":[

-0.65

stad

-0.65

!/

-0.65

Jr

-0.63

POSITIVE LOGITS

own

1.64

 predecessor

1.04

 namesake

1.03

Own

1.01

 flagship

0.92

ELF

0.92

 entirety

0.92

 newest

0.90

 tentacles

0.84

 latest

0.83

Activations Density 0.131%

pronouns and possessive adjectives

No Comments

No Known Activations