INDEX

Explanations

phrases or words indicating a negative evaluation or criticism

oai_token-act-pair · gpt-3.5-turbo

expressions of dissatisfaction or negativity

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

 Greenwood

-0.86

Arthur

-0.76

 Arthur

-0.74

 Armory

-0.73

 Clarke

-0.69

 Expression

-0.67

kar

-0.66

Ard

-0.66

 Hampton

-0.66

 Manor

-0.66

POSITIVE LOGITS

 suck

3.70

 sucks

2.76

 sucked

2.71

 sucking

2.38

 fucked

1.41

 sucker

1.40

 fuck

1.37

 stink

1.37

 siph

1.27

 screwed

1.21

Activations Density 0.017%

phrases or words indicating a negative evaluation or criticism

expressions of dissatisfaction or negativity

No Comments

No Known Activations