INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

situations involving violence or crime

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

theless

-0.67

 anecd

-0.63

 unequivocally

-0.59

ulative

-0.58

 underestimated

-0.58

 learnt

-0.56

 fundamentally

-0.54

 succeeded

-0.53

(>

-0.52

 logically

-0.51

POSITIVE LOGITS

 picnic

0.65

 kios

0.63

 dinner

0.63

 mansion

0.62

 encamp

0.61

 mast

0.60

 supper

0.59

 tower

0.58

 cafe

0.58

 airst

0.58

Activations Density 1.071%

locations or places

situations involving violence or crime

No Comments

No Known Activations