INDEX

Explanations

phrases related to additional information or actions that follow an initial event or statement

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "follow up."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

tu

-0.73

enberg

-0.72

pload

-0.68

 magnet

-0.68

¬¼

-0.66

mir

-0.63

DI

-0.61

 projecting

-0.60

erved

-0.60

 ORIG

-0.60

POSITIVE LOGITS

ounter

0.74

 generations

0.73

alogue

0.73

orate

0.73

rocal

0.72

 developments

0.70

adesh

0.70

Redd

0.70

 coli

0.70

ilee

0.68

Activations Density 0.029%

phrases related to additional information or actions that follow an initial event or statement

instances of the phrase "follow up."

No Comments

No Known Activations