INDEX

Explanations

phrases indicating the purpose or intention behind specific actions or decisions

oai_token-act-pair · gpt-3.5-turbo

phrases indicating intent or purpose

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Gone

-0.70

typ

-0.69

 Appears

-0.69

listed

-0.67

done

-0.67

 Warning

-0.66

 Orange

-0.65

 Verse

-0.64

raped

-0.64

 Steps

-0.63

POSITIVE LOGITS

 maximize

1.47

 achieve

1.43

 minimize

1.36

 fulfill

1.35

 facilitate

1.33

 satisfy

1.32

 preserve

1.29

 simplify

1.29

 obtain

1.29

 accommodate

1.29

Activations Density 0.119%

phrases indicating the purpose or intention behind specific actions or decisions

phrases indicating intent or purpose

No Comments

No Known Activations