INDEX

Explanations

the word "assume" at varying strength levels

oai_token-act-pair · gpt-3.5-turbo

the word "assume" and its variations in the context of making assumptions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Rebels

-0.64

stre

-0.61

 Comm

-0.59

hod

-0.58

 worthy

-0.58

 Christmas

-0.58

zan

-0.57

 reson

-0.57

Bee

-0.57

 Wars

-0.56

POSITIVE LOGITS

 assume

3.73

 assumes

2.50

 presume

2.43

 assumed

2.12

 assuming

1.97

 suppose

1.74

 assum

1.72

 assumption

1.69

 imply

1.48

 conclude

1.44

Activations Density 0.013%

the word "assume" at varying strength levels

the word "assume" and its variations in the context of making assumptions

No Comments

No Known Activations