INDEX

Explanations

verbs indicating beliefs or opinions

oai_token-act-pair · gpt-3.5-turbo

expressions of belief or conviction

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

grad

-0.71

bourg

-0.61

ward

-0.60

arter

-0.57

 Strikes

-0.57

 encro

-0.56

 Rolls

-0.55

 Siege

-0.55

 Scourge

-0.54

 intens

-0.54

POSITIVE LOGITS

 believe

3.33

 Believe

2.21

 believes

2.11

 believing

1.93

 believed

1.84

 belief

1.56

 think

1.47

 presume

1.37

BEL

1.33

 believer

1.33

Activations Density 0.025%

verbs indicating beliefs or opinions

expressions of belief or conviction

No Comments

No Known Activations