INDEX

Explanations

past habitual actions or states

oai_token-act-pair · gpt-3.5-turbo

the phrase "used to" indicating past habits or states

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

edIn

-0.76

rising

-0.74

leaf

-0.74

oval

-0.73

imov

-0.69

agonists

-0.68

irez

-0.68

states

-0.67

IDA

-0.66

fixes

-0.65

POSITIVE LOGITS

 joke

1.16

 haunt

1.01

 rely

1.01

be

1.00

 tease

0.97

 adore

0.95

 operate

0.95

 laugh

0.94

 belong

0.94

 dominate

0.94

Activations Density 0.027%

past habitual actions or states

the phrase "used to" indicating past habits or states

No Comments

No Known Activations