INDEX

Explanations

terms related to enclosed spaces within a building, particularly hallways and doorways

oai_token-act-pair · gpt-3.5-turbo

references to physical spaces and conflicts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ament

-0.81

acles

-0.80

sonian

-0.79

estones

-0.76

estone

-0.75

nets

-0.75

netic

-0.73

iments

-0.72

ifi

-0.69

net

-0.69

POSITIVE LOGITS

xual

0.77

[|

0.60

 PRESIDENT

0.59

 Hundred

0.58

etheless

0.54

 recalling

0.54

 hesitate

0.52

 interrupted

0.52

cellence

0.52

armac

0.51

Activations Density 0.075%

terms related to enclosed spaces within a building, particularly hallways and doorways

references to physical spaces and conflicts

No Comments

No Known Activations