INDEX

Explanations

phrases related to commands or instructions

oai_token-act-pair · gpt-3.5-turbo

references to personal experiences or possessions

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

 Rhodes

-0.73

 Hercules

-0.70

 Kissinger

-0.65

 Gork

-0.61

 Sussex

-0.60

ktop

-0.59

 Nelson

-0.58

 Lange

-0.58

 Chavez

-0.57

 Wellington

-0.57

POSITIVE LOGITS

âĢ

2.56

 âĢ

2.13

ãĢ

1.72

Â¨

1.56

âĻ¦

1.52

âľ

1.52

âĶ

1.52

âĢł

1.49

âĸł

1.49

âĹ

1.48

Activations Density 1.461%

phrases related to commands or instructions

references to personal experiences or possessions

No Comments

No Known Activations