INDEX

Explanations

instances of the word "let" followed by a high activation value of 9 or 10

oai_token-act-pair · gpt-3.5-turbo

instances of the word "let."

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

 Heroic

-0.64

 results

-0.61

 seats

-0.61

 Dark

-0.61

 proprietary

-0.58

 dead

-0.58

 students

-0.57

 offices

-0.55

 advisors

-0.55

 Universe

-0.55

POSITIVE LOGITS

let

4.61

lets

3.75

LET

2.78

letes

1.95

lete

1.93

lette

1.84

lett

1.78

leted

1.49

lest

1.45

leton

1.34

Activations Density 0.020%

instances of the word "let" followed by a high activation value of 9 or 10

instances of the word "let."

No Comments

No Known Activations