INDEX

Explanations

words related to geographical locations, specifically focusing on Cambridge

oai_token-act-pair · gpt-3.5-turbo

mentions of the city Cambridge

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

ued

-0.74

ore

-0.67

 idols

-0.67

 lava

-0.65

 Romo

-0.65

upd

-0.64

lol

-0.64

 idol

-0.64

 Nightmare

-0.63

jer

-0.62

POSITIVE LOGITS

 Cambridge

3.74

 Oxford

2.07

 Harvard

1.73

 Princeton

1.66

 Worcester

1.62

Cam

1.59

 Copenhagen

1.57

 Waterloo

1.48

 Plymouth

1.47

 Sussex

1.45

Activations Density 0.015%

words related to geographical locations, specifically focusing on Cambridge

mentions of the city Cambridge

No Comments

No Known Activations