INDEX

Explanations

negations indicating something is "not" happening or being done

oai_token-act-pair · gpt-3.5-turbo

negations or expressions of noncompliance

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

¿½

-0.86

soType

-0.75

raltar

-0.74

 Pric

-0.68

ãĤ¦ãĤ¹

-0.67

 Done

-0.67

 Palest

-0.66

ģĸ

-0.65

 Parenthood

-0.65

 Maiden

-0.65

POSITIVE LOGITS

 bother

1.27

 hesitate

1.26

 respond

1.23

 realize

1.18

 realise

1.15

 want

1.13

 participate

1.13

 recognize

1.13

 receive

1.12

 approve

1.11

Activations Density 0.099%

negations indicating something is "not" happening or being done

negations or expressions of noncompliance

No Comments

No Known Activations