INDEX

Explanations

statements indicating that more information can be found elsewhere

oai_token-act-pair · gpt-3.5-turbo

references to completeness or entirety in information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

onis

-0.69

fu

-0.68

rosso

-0.68

_-

-0.66

 Fever

-0.65

zy

-0.64

fw

-0.64

yne

-0.61

wo

-0.61

Ear

-0.60

POSITIVE LOGITS

 brunt

0.93

 extent

0.83

 underside

0.80

 impetus

0.78

 embodiment

0.78

 alphabet

0.78

 wording

0.75

 nature

0.74

 spectrum

0.73

 fullest

0.71

Activations Density 0.099%

statements indicating that more information can be found elsewhere

references to completeness or entirety in information

No Comments

No Known Activations