INDEX

Explanations

phrases related to different kinds of reviews

oai_token-act-pair · gpt-3.5-turbo

references to reviews

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

plex

-0.70

nown

-0.70

ata

-0.69

htar

-0.68

atum

-0.67

Sac

-0.66

 forcibly

-0.65

stroke

-0.63

oshi

-0.63

ossus

-0.63

POSITIVE LOGITS

 reviews

3.95

 Reviews

2.67

 review

2.46

 reviewers

2.41

review

2.35

 reviewer

2.14

Review

2.03

 Review

2.02

 reviewed

1.82

reviewed

1.80

Activations Density 0.012%

phrases related to different kinds of reviews

references to reviews

No Comments

No Known Activations