INDEX

Explanations

phrases related to starting or initiating an action

oai_token-act-pair · gpt-3.5-turbo

verbs indicating the initiation or continuation of actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Surprise

-0.69

 compliment

-0.66

contin

-0.64

fortunately

-0.64

 continuation

-0.63

astern

-0.62

prem

-0.61

sun

-0.60

fore

-0.60

lat

-0.58

POSITIVE LOGITS

 anew

0.74

ories

0.70

ORK

0.68

edIn

0.67

ixel

0.66

phas

0.65

hao

0.63

 circuits

0.61

daq

0.61

ury

0.60

Activations Density 0.153%

phrases related to starting or initiating an action

verbs indicating the initiation or continuation of actions

No Comments

No Known Activations