INDEX

Explanations

phrases related to actions or instructions

oai_token-act-pair · gpt-3.5-turbo

symbols or characters that appear repeatedly

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Donna

-0.73

 Billy

-0.72

 disse

-0.71

 dist

-0.69

 laundry

-0.68

 unbeliev

-0.68

 Harley

-0.68

DeL

-0.67

 Miss

-0.66

Haj

-0.66

POSITIVE LOGITS

ĺ

1.77

ĺħ

0.98

right

0.95

Ĳ

0.94

ĸ

0.92

Ð¾

0.92

uo

0.89

rax

0.89

ĥ

0.89

Ĺ

0.88

Activations Density 0.082%

phrases related to actions or instructions

symbols or characters that appear repeatedly

No Comments

No Known Activations