INDEX

Explanations

dates and numbers

oai_token-act-pair · gpt-3.5-turbo

statements of curiosity and exploration

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

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

ALL

-0.92

 EVERY

-0.90

 goddamn

-0.74

!!!

-0.71

!!!!

-0.71

ALL

-0.69

 Absolutely

-0.69

.#

-0.68

 AMERICA

-0.68

========

-0.67

POSITIVE LOGITS

artney

0.71

 quicker

0.65

 softer

0.64

 recons

0.62

 neighbouring

0.62

 later

0.62

 slower

0.60

 earlier

0.60

 sharper

0.60

 nearer

0.59

Activations Density 1.043%

dates and numbers

statements of curiosity and exploration

No Comments

No Known Activations