INDEX

Explanations

phrases indicating contrasting information

oai_token-act-pair · gpt-3.5-turbo

phrases that begin with "Yet."

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

 starting

-0.63

 basically

-0.60

 interests

-0.59

 balls

-0.59

 grav

-0.56

 orders

-0.56

 comm

-0.56

 advis

-0.56

 kind

-0.54

 periodically

-0.54

POSITIVE LOGITS

Yet

3.21

Yet

2.72

yet

2.53

yet

2.09

 Nonetheless

1.51

 Nevertheless

1.48

But

1.44

 Indeed

1.40

 Worse

1.34

 Moreover

1.32

Activations Density 0.018%

phrases indicating contrasting information

phrases that begin with "Yet."

No Comments

No Known Activations