INDEX

Explanations

references to organizations, events or places

oai_token-act-pair · gpt-3.5-turbo

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

é¾į

-0.73

Ł

-0.72

Ķ

-0.70

uci

-0.70

role

-0.69

TeX

-0.67

]=

-0.67

ibo

-0.67

Throw

-0.66

rase

-0.66

POSITIVE LOGITS

 however

0.73

 meanwhile

0.63

 Schn

0.63

Gil

0.62

Kap

0.62

 Goldstein

0.62

 though

0.60

atever

0.59

Gus

0.59

huh

0.58

Activations Density 0.318%

references to organizations, events or places

No Comments

No Known Activations