INDEX

Explanations

proper nouns or names mentioned in text

oai_token-act-pair · gpt-3.5-turbo

references to names and identities

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

eger

-0.72

ardless

-0.71

nesday

-0.70

pell

-0.69

rising

-0.69

riott

-0.68

pelling

-0.68

chet

-0.67

odic

-0.65

 guiName

-0.65

POSITIVE LOGITS

 spans

0.77

 resides

0.73

 resembled

0.72

çīĪ

0.72

 exploits

0.71

 resembles

0.70

 coincides

0.69

 clocks

0.69

 derives

0.69

 overl

0.68

Activations Density 0.069%

proper nouns or names mentioned in text

references to names and identities

No Comments

No Known Activations