INDEX

Explanations

words related to original or source material

oai_token-act-pair · gpt-3.5-turbo

words related to origin and indoctrination

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 squared

-0.66

 plaus

-0.60

SPONSORED

-0.58

#$

-0.57

waves

-0.55

 alternatively

-0.55

farm

-0.54

 patterns

-0.54

note

-0.54

 ende

-0.54

POSITIVE LOGITS

inated

2.57

inating

2.55

ination

2.54

inate

2.40

inations

2.29

inates

2.28

inator

2.16

inators

2.14

ined

1.93

inals

1.89

Activations Density 0.116%

words related to original or source material

words related to origin and indoctrination

No Comments

No Known Activations