INDEX

Explanations

mentions of things that are slightly different or slightly more/less than something else

oai_token-act-pair · gpt-3.5-turbo

the word "slightly" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

elsen

-0.75

 STATS

-0.74

ravings

-0.69

 Aviv

-0.68

vre

-0.66

itivity

-0.65

rights

-0.65

ilee

-0.64

inarily

-0.64

ulkan

-0.64

POSITIVE LOGITS

 altered

0.93

 misleading

0.86

 eccentric

0.85

 tilted

0.83

 differently

0.83

 tweaked

0.82

 modified

0.81

 inconven

0.81

 inflated

0.80

 biased

0.80

Activations Density 0.035%

mentions of things that are slightly different or slightly more/less than something else

the word "slightly" and its variations

No Comments

No Known Activations