INDEX

Explanations

specific company names

oai_token-act-pair · gpt-3.5-turbo

brands and company names related to products or retailers

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

 bullpen

-0.70

................................................................

-0.69

ball

-0.69

 defense

-0.68

Washington

-0.67

cgi

-0.64

 characterize

-0.62

 Indians

-0.62

ALL

-0.61

 characterized

-0.61

POSITIVE LOGITS

Tes

4.10

Tes

3.63

Kes

1.34

Cad

1.20

 Lanc

1.20

 Aber

1.15

 Marks

1.13

Tos

1.12

 Mond

1.12

 Nvidia

1.08

Activations Density 0.030%

specific company names

brands and company names related to products or retailers

No Comments

No Known Activations