INDEX

Explanations

phrases describing memories or recollections

oai_token-act-pair · gpt-3.5-turbo

instances of the word "recalled."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

MIT

-0.65

 crypto

-0.62

cept

-0.61

 Patreon

-0.60

 Trop

-0.59

ÏĢ

-0.59

Crypt

-0.58

cul

-0.58

oons

-0.58

atures

-0.57

POSITIVE LOGITS

 recalled

3.94

 recalls

2.78

 recall

2.49

 recalling

2.43

 remembered

2.02

 recounted

1.96

 remembers

1.87

 Recall

1.70

 recounts

1.52

 recol

1.47

Activations Density 0.022%

phrases describing memories or recollections

instances of the word "recalled."

No Comments

No Known Activations