INDEX

Explanations

references to achieving success and successful outcomes

oai_token-act-pair · gpt-3.5-turbo

expressions related to success and achievement

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Origin

-0.75

collection

-0.72

origin

-0.69

view

-0.68

uns

-0.67

Origin

-0.66

pled

-0.66

picking

-0.65

 Geneva

-0.64

Americ

-0.64

POSITIVE LOGITS

 succeed

1.27

 succeeds

1.21

 succeeded

0.90

anke

0.84

 Pengu

0.82

SourceFile

0.80

rupal

0.79

nir

0.78

 neighb

0.77

 mathemat

0.75

Activations Density 0.009%

references to achieving success and successful outcomes

expressions related to success and achievement

No Comments

No Known Activations