INDEX

Explanations

phrases related to authority and control

oai_token-act-pair · gpt-3.5-turbo

phrases indicating denial or negation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 adolesc

-0.69

CTV

-0.63

 compat

-0.61

rossover

-0.61

rencies

-0.60

IFE

-0.58

atoes

-0.58

 transformations

-0.58

ricular

-0.58

 speculation

-0.58

POSITIVE LOGITS

 intervene

1.27

 relent

1.01

 approve

1.00

 sanction

0.99

 evict

0.97

 incentiv

0.97

 swoop

0.95

 dictate

0.95

 forbid

0.91

 veto

0.91

Activations Density 0.668%

phrases related to authority and control

phrases indicating denial or negation

No Comments

No Known Activations