INDEX

Explanations

references to locations or historical events

oai_token-act-pair · gpt-3.5-turbo

references to historical groups or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

panic

-0.87

erate

-0.80

ware

-0.74

osate

-0.73

abal

-0.73

ulative

-0.72

 solder

-0.71

vous

-0.70

milo

-0.69

ieu

-0.68

POSITIVE LOGITS

oken

0.64

Ab

0.63

âĹ¼

0.63

ãĤ´ãĥ³

0.63

 Holt

0.62

 circum

0.62

Dec

0.61

_.

0.61

Liter

0.60

isd

0.59

Activations Density 0.000%

references to locations or historical events

references to historical groups or entities

No Comments

No Known Activations