INDEX

Explanations

contrary statements about potential outcomes

oai_token-act-pair · gpt-3.5-turbo

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

iddle

-0.63

raft

-0.62

iling

-0.59

ding

-0.58

ature

-0.58

 Enhance

-0.58

rift

-0.57

ishing

-0.55

anwhile

-0.54

 guiName

-0.53

POSITIVE LOGITS

 liked

1.12

 been

1.06

 gotten

1.02

 benefited

0.99

been

0.98

 preferred

0.96

 fared

0.94

 gladly

0.91

 gone

0.85

 avoided

0.85

Activations Density 0.054%

contrary statements about potential outcomes

No Comments

No Known Activations