INDEX

Explanations

specific technical features or attributes indicated by numerical values

oai_token-act-pair · gpt-3.5-turbo

possessive verbs indicating ownership or attributes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

TG

-0.66

Interested

-0.65

 endeavour

-0.64

FW

-0.63

iss

-0.62

iling

-0.60

ensing

-0.59

YING

-0.59

MO

-0.58

fty

-0.58

POSITIVE LOGITS

 been

1.57

 undergone

1.46

been

1.32

 Been

1.17

 become

1.14

 survived

1.12

 existed

1.12

 gotten

1.06

stood

1.06

 arisen

1.04

Activations Density 0.187%

specific technical features or attributes indicated by numerical values

possessive verbs indicating ownership or attributes

No Comments

No Known Activations