INDEX

Explanations

dates in the format month/day/year that are mentioned in text

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Posted" followed by numbers, indicating publication dates or timestamps

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ãĥ¯ãĥ³

-0.81

oller

-0.67

ifted

-0.63

ingly

-0.63

stood

-0.58

negie

-0.57

 pains

-0.53

 Weinstein

-0.53

 consciousness

-0.52

wit

-0.52

POSITIVE LOGITS

0.97

0.92

Apr

0.92

0.91

Thu

0.91

0.90

0.88

Tue

0.87

Activations Density 0.021%

dates in the format month/day/year that are mentioned in text

instances of the word "Posted" followed by numbers, indicating publication dates or timestamps

No Comments

No Known Activations