INDEX

Explanations

mentions of numbers or measurements

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific punctuation symbols or brackets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

rette

-0.72

ĵĺ

-0.68

aband

-0.67

 neigh

-0.66

tones

-0.66

 cradle

-0.65

marine

-0.65

etsy

-0.65

ãĤ©

-0.64

reet

-0.64

POSITIVE LOGITS

 However

0.96

 Additionally

0.91

 Furthermore

0.85

 Similarly

0.84

 Conversely

0.82

 Likewise

0.82

 Later

0.81

 Nevertheless

0.80

 Therefore

0.77

 Alternatively

0.76

Activations Density 0.050%

mentions of numbers or measurements

occurrences of specific punctuation symbols or brackets

No Comments

No Known Activations