INDEX

Explanations

the word "re" in texts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "we" and other variations of "to be" in connected contexts, indicating collective experiences or observations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Site

-0.61

ãĥ©ãĥ³

-0.57

mx

-0.53

Spot

-0.53

ãĤµ

-0.52

****************

-0.51

Search

-0.51

SEE

-0.49

 SUPER

-0.49

Indust

-0.49

POSITIVE LOGITS

uddenly

0.68

fw

0.66

gyn

0.65

helle

0.63

agh

0.63

lynn

0.63

ensable

0.62

addafi

0.61

hew

0.61

eal

0.60

Activations Density 0.613%

the word "re" in texts

instances of the word "we" and other variations of "to be" in connected contexts, indicating collective experiences or observations

No Comments

No Known Activations