INDEX

Explanations

words related to the concept of elections being manipulated or biased

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

torch.float32

Hook Point

blocks.8.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

anse

-0.75

chal

-0.71

lication

-0.69

phrine

-0.68

udi

-0.67

 Spiegel

-0.67

esa

-0.66

antry

-0.66

ijk

-0.65

ppard

-0.64

POSITIVE LOGITS

 rigged

1.33

 rigging

1.05

rig

0.89

ument

0.81

 against

0.75

 rotten

0.74

eering

0.73

 cages

0.71

lock

0.68

 twist

0.67

Activations Density 0.009%

words related to the concept of elections being manipulated or biased

No Comments

No Known Activations