INDEX

Explanations

names related to "Lor" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to the term "lor."

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

 Startup

-0.79

 Saber

-0.65

 Sprint

-0.64

 satellite

-0.62

 Sagan

-0.62

 startups

-0.61

 Scouts

-0.60

 Siri

-0.60

 navig

-0.60

 Push

-0.60

POSITIVE LOGITS

lor

4.57

lore

1.86

lam

1.35

la

1.15

los

1.09

lah

1.09

lon

1.08

rolet

0.99

liv

0.98

lez

0.97

Activations Density 0.011%

names related to "Lor" in various contexts

references to the term "lor."

No Comments

No Known Activations