INDEX

Explanations

names and specific information about individuals

oai_token-act-pair · gpt-3.5-turbo

references to specific individuals and their actions

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

æ©Ł

-0.75

ãĥĦ

-0.66

DCS

-0.64

Ħ¢

-0.59

ãĤ±

-0.54

wcsstore

-0.54

)?

-0.52

å¥

-0.51

ãĥĥãĥĪ

-0.51

ãĤ«

-0.51

POSITIVE LOGITS

ecause

0.61

 though

0.57

 whereas

0.53

 however

0.52

but

0.52

 because

0.51

[];

0.50

(>

0.48

 Ezek

0.47

Activations Density 3.727%

names and specific information about individuals

references to specific individuals and their actions

No Comments

No Known Activations