INDEX

Explanations

proper nouns related to the brand "Whole"

oai_token-act-pair · gpt-3.5-turbo

references to the term "Whole" and its related contexts

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

 suicide

-0.69

isphere

-0.69

isi

-0.67

 idol

-0.67

igan

-0.66

 goblins

-0.65

 regression

-0.65

 recogn

-0.64

 draw

-0.64

 laure

-0.63

POSITIVE LOGITS

 Whole

3.66

 Trader

1.35

 Healthy

1.24

 Minute

1.17

 Taste

1.15

 Straight

1.14

 Grain

1.13

 Fresh

1.11

 Target

1.08

 Thick

1.02

Activations Density 0.027%

proper nouns related to the brand "Whole"

references to the term "Whole" and its related contexts

No Comments

No Known Activations