INDEX

Explanations

phrases related to official processes and document reviews

oai_token-act-pair · gpt-3.5-turbo

terms related to regulatory processes and formal approvals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 clenched

-0.73

 suspic

-0.69

 Tale

-0.63

 metic

-0.61

 nutshell

-0.61

olver

-0.61

 advoc

-0.61

 compr

-0.60

 corrid

-0.59

rame

-0.59

POSITIVE LOGITS

 purposes

1.24

ability

1.10

 elsewhere

0.96

dates

0.88

tests

0.84

 anywhere

0.83

worthiness

0.83

BILITY

0.81

 status

0.81

 consideration

0.81

Activations Density 0.224%

phrases related to official processes and document reviews

terms related to regulatory processes and formal approvals

No Comments

No Known Activations