INDEX

Explanations

resources or information that can be accessed or located

oai_token-act-pair · gpt-3.5-turbo

phrases indicating availability or the potential to access information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

terday

-0.71

 Newspaper

-0.63

lights

-0.62

 Matters

-0.59

pour

-0.57

 Maver

-0.57

 Pear

-0.56

 Cipher

-0.56

 Soon

-0.55

Cry

-0.55

POSITIVE LOGITS

 traced

1.22

 viewed

1.16

 seen

1.11

 accessed

1.11

 summed

1.09

 summarized

1.07

 found

1.06

 construed

0.99

 downloaded

0.98

 inferred

0.96

Activations Density 0.055%

resources or information that can be accessed or located

phrases indicating availability or the potential to access information

No Comments

No Known Activations