INDEX

Explanations

phrases related to the act of playing sports

oai_token-act-pair · gpt-3.5-turbo

references to activities or roles in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

emetery

-0.66

ilts

-0.62

ablishment

-0.62

¿½

-0.62

 Leilan

-0.61

ortium

-0.60

Delete

-0.60

IDS

-0.59

 shipments

-0.58

 nominations

-0.58

POSITIVE LOGITS

 Mahjong

0.80

arro

0.73

 drums

0.73

 symp

0.71

natureconservancy

0.66

ãĤ°

0.65

 role

0.63

 darn

0.62

 differently

0.61

cha

0.60

Activations Density 0.174%

phrases related to the act of playing sports

references to activities or roles in various contexts

No Comments

No Known Activations