INDEX

Explanations

words related to medical conditions, particularly those ending in "-cul"

oai_token-act-pair · gpt-3.5-turbo

words related to calculations and mathematical concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

bye

-0.69

 snaps

-0.68

 deal

-0.68

 Benny

-0.66

iso

-0.65

TD

-0.64

 hots

-0.63

 leash

-0.63

 Sunshine

-0.62

 swift

-0.61

POSITIVE LOGITS

cul

4.56

culated

2.92

culus

2.82

culation

2.26

cular

1.91

cu

1.40

cule

1.38

cules

1.35

cus

1.32

cium

1.21

Activations Density 0.016%

words related to medical conditions, particularly those ending in "-cul"

words related to calculations and mathematical concepts

No Comments

No Known Activations