INDEX

Explanations

phrases related to accessibility

oai_token-act-pair · gpt-3.5-turbo

references to "access."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

stal

-0.77

pload

-0.71

 Kern

-0.64

 Sabha

-0.63

 sperm

-0.61

izarre

-0.61

 Fenrir

-0.60

 ejac

-0.60

hovah

-0.59

 autonom

-0.59

POSITIVE LOGITS

ibility

1.49

ible

1.19

ibly

1.16

ibles

1.07

ory

1.01

IBLE

0.96

ibl

0.89

ibilities

0.88

ions

0.81

âĶģ

0.81

Activations Density 0.020%

phrases related to accessibility

references to "access."

No Comments

No Known Activations