INDEX

Explanations

phrases or sentences with the structured format "then-number"

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 ticks

-0.76

 melt

-0.72

 flex

-0.70

 weave

-0.69

 rede

-0.69

 amplification

-0.67

 affected

-0.67

 reset

-0.67

 theorem

-0.66

 wipes

-0.66

POSITIVE LOGITS

cand

1.40

president

1.34

Senator

1.27

President

1.21

Prime

1.17

current

1.16

Govern

1.16

director

1.16

Mayor

1.14

recent

1.11

Activations Density 0.032%

phrases or sentences with the structured format "then-number"

No Comments

No Known Activations