INDEX

Explanations

action words related to making decisions or taking steps

oai_token-act-pair · gpt-3.5-turbo

actions related to initiating movement or making decisions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

atched

-0.73

icons

-0.73

nesota

-0.71

roximately

-0.70

wo

-0.68

alde

-0.63

terday

-0.62

Staff

-0.61

WARE

-0.61

ants

-0.59

POSITIVE LOGITS

 drastic

0.94

 unorthodox

0.86

 risky

0.82

 alternatives

0.82

 unconventional

0.79

 alternative

0.78

odox

0.78

 gamble

0.77

 aggressive

0.76

 plunge

0.75

Activations Density 0.340%

action words related to making decisions or taking steps

actions related to initiating movement or making decisions

No Comments

No Known Activations