INDEX

Explanations

tech companies and geopolitical entities

oai_token-act-pair · gpt-3.5-turbo

instances of commas indicating lists or separations between multiple items or concepts

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

uces

-0.76

-0.69

worldly

-0.68

 Starts

-0.65

,—

-0.64

EngineDebug

-0.60

(>

-0.57

esi

-0.55

Enlarge

-0.54

!/

-0.54

POSITIVE LOGITS

are

1.22

 were

1.15

were

1.14

have

1.08

 aren

1.04

 have

1.04

 weren

1.00

 respectively

0.98

 appear

0.96

 constitute

0.94

Activations Density 0.148%

tech companies and geopolitical entities

instances of commas indicating lists or separations between multiple items or concepts

No Comments

No Known Activations