INDEX

Explanations

names or titles in quotation marks

oai_token-act-pair · gpt-3.5-turbo

quotation marks or speech indicators

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

 reconcil

-0.76

 appointments

-0.73

 repay

-0.70

 populated

-0.70

 peripheral

-0.70

 menstrual

-0.69

 replication

-0.68

 aggregate

-0.67

 childbirth

-0.66

 furn

-0.66

POSITIVE LOGITS

Fat

1.17

Skip

1.11

Big

1.06

Rust

1.05

Golden

1.04

Spirit

1.04

Fuck

1.04

Bob

1.02

Bull

1.02

Hope

1.00

Activations Density 0.077%

names or titles in quotation marks

quotation marks or speech indicators

No Comments

No Known Activations