INDEX

Explanations

instances where instructions or expectations are being set or emphasized

oai_token-act-pair · gpt-3.5-turbo

" a" in statements about being something, eg "I am a" or "You can be a"

Explanation Uploaded by User

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 interfaces

-0.71

Events

-0.69

IDs

-0.68

uden

-0.68

mares

-0.66

 appointments

-0.65

 assemblies

-0.65

 assessments

-0.64

 analogous

-0.64

onies

-0.64

POSITIVE LOGITS

 believer

1.16

 member

1.09

 hypocr

1.07

verse

1.02

 participant

1.00

 supporter

0.99

 proponent

0.99

 spectator

0.98

 follower

0.95

 beneficiary

0.93

Activations Density 0.169%

instances where instructions or expectations are being set or emphasized

" a" in statements about being something, eg "I am a" or "You can be a"

No Comments

No Known Activations