INDEX

Explanations

references to specific cases or situations

oai_token-act-pair · gpt-3.5-turbo

instances of the term "Case" in various contexts

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

 overwhelm

-0.73

 glory

-0.66

 handic

-0.65

pless

-0.64

 turb

-0.64

 leisure

-0.62

gob

-0.61

 obey

-0.60

 surfing

-0.60

 drunk

-0.59

POSITIVE LOGITS

 Case

4.01

Case

2.70

 Cases

2.56

 CASE

2.42

case

2.16

 case

2.13

cases

1.80

 cases

1.60

 Casey

1.10

 Matter

0.96

Activations Density 0.009%

references to specific cases or situations

instances of the term "Case" in various contexts

No Comments

No Known Activations