INDEX

Explanations

conditional statements expressing different possibilities

oai_token-act-pair · gpt-3.5-turbo

phrases indicating uncertainty or conditional situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 unlikely

-0.65

é¾įå¥ĳå£«

-0.65

©¶æ¥µ

-0.64

 unanswered

-0.63

vier

-0.60

ilib

-0.58

NRS

-0.57

ãĢĲ

-0.57

 Registry

-0.56

 nowhere

-0.55

POSITIVE LOGITS

 consciously

0.86

 originate

0.74

be

0.74

 succeeds

0.70

 succeed

0.68

 intentional

0.68

 coasts

0.68

cause

0.66

 choose

0.66

ween

0.62

Activations Density 0.066%

conditional statements expressing different possibilities

phrases indicating uncertainty or conditional situations

No Comments

No Known Activations