INDEX

Explanations

words related to specific names and locations, particularly those related to technology or cultural terms such as Kalashnikov

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

catentry

-0.98

ascript

-0.84

ilial

-0.79

 Malf

-0.69

tyard

-0.68

 Qiao

-0.67

 SERV

-0.67

Versions

-0.67

UCT

-0.65

essions

-0.63

POSITIVE LOGITS

 Lumpur

1.09

EStreamFrame

0.90

utra

0.85

EStream

0.83

istan

0.81

enei

0.81

schild

0.79

bye

0.77

ikuman

0.74

gaard

0.74

Activations Density 0.411%

words related to specific names and locations, particularly those related to technology or cultural terms such as Kalashnikov

references to specific locations or entities

No Comments

No Known Activations