INDEX

Explanations

phrases related to physical actions or descriptions

oai_token-act-pair · gpt-3.5-turbo

pronouns related to people

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Gerard

-0.69

Mit

-0.67

 Majority

-0.66

 Emer

-0.65

 Cliff

-0.64

 Pompe

-0.64

 Beacon

-0.64

 Junk

-0.64

 Anat

-0.63

 Petraeus

-0.63

POSITIVE LOGITS

arers

0.98

pton

0.91

ÃĥÃĤ

0.91

ared

0.90

've

0.90

arer

0.89

til

0.88

'll

0.86

hots

0.85

ldon

0.84

Activations Density 0.305%

phrases related to physical actions or descriptions

pronouns related to people

No Comments

No Known Activations