INDEX

Explanations

surrounding words related to someone or a group being upset

oai_token-act-pair · gpt-3.5-turbo

instances of the word "upset" and related expressions of emotional distress

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

Base

-0.73

basic

-0.67

Pract

-0.66

 Spiegel

-0.66

Mand

-0.65

Compan

-0.65

cd

-0.65

 Protocol

-0.64

 Membership

-0.63

lder

-0.61

POSITIVE LOGITS

 upset

3.71

 offended

1.85

 angered

1.75

 upsetting

1.70

 annoyed

1.68

 irritated

1.62

 pissed

1.61

 outraged

1.61

 enraged

1.53

 embarrassed

1.50

Activations Density 0.017%

surrounding words related to someone or a group being upset

instances of the word "upset" and related expressions of emotional distress

No Comments

No Known Activations