INDEX

Explanations

phrases indicating a condition, situation, or problem

oai_token-act-pair · gpt-3.5-turbo

mention of potential legal or criminal issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

çīĪ

-0.85

©¶æ¥µ

-0.78

ãĤ´ãĥ³

-0.77

ãĤ¤ãĥĪ

-0.75

 srfAttach

-0.73

folios

-0.69

 Moroc

-0.69

renheit

-0.68

ÃįÃį

-0.68

ô

-0.67

POSITIVE LOGITS

 sufficiently

0.84

pires

0.82

 somebody

0.79

 disrespect

0.76

 truly

0.75

 slightest

0.74

 properly

0.73

 ever

0.71

 fails

0.69

 sake

0.68

Activations Density 0.292%

phrases indicating a condition, situation, or problem

mention of potential legal or criminal issues

No Comments

No Known Activations