Notes
-----
- The function assumes that the input quaternions have unit norm.
- It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
+ The function assumes that the input quaternions have unit norm. It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
"""
(
a_h,
@@ -471,8 +470,7 @@ Source code for dicee.models.quaternion
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
- formed with these head entities and relations against a subset of entities, i.e.,
- [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+ formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the `target_entity_idx`, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/_sources/autoapi/dicee/index.rst.txt b/_sources/autoapi/dicee/index.rst.txt
index 820b52b1..b6a26ec1 100644
--- a/_sources/autoapi/dicee/index.rst.txt
+++ b/_sources/autoapi/dicee/index.rst.txt
@@ -2358,8 +2358,7 @@ Attributes
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
- formed with these head entities and relations against a subset of entities, i.e.,
- [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+ formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the `target_entity_idx`, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/_sources/autoapi/dicee/models/index.rst.txt b/_sources/autoapi/dicee/models/index.rst.txt
index 3846f629..d9c855e8 100644
--- a/_sources/autoapi/dicee/models/index.rst.txt
+++ b/_sources/autoapi/dicee/models/index.rst.txt
@@ -1818,8 +1818,7 @@ Functions
.. rubric:: Notes
- The function assumes that the input quaternions have unit norm.
- It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
+ The function assumes that the input quaternions have unit norm. It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
.. py:class:: QMult(args: dict)
@@ -1996,8 +1995,7 @@ Functions
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
- formed with these head entities and relations against a subset of entities, i.e.,
- [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+ formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the `target_entity_idx`, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/_sources/autoapi/dicee/models/quaternion/index.rst.txt b/_sources/autoapi/dicee/models/quaternion/index.rst.txt
index c29f3b08..7b67977f 100644
--- a/_sources/autoapi/dicee/models/quaternion/index.rst.txt
+++ b/_sources/autoapi/dicee/models/quaternion/index.rst.txt
@@ -41,8 +41,7 @@ Functions
.. rubric:: Notes
- The function assumes that the input quaternions have unit norm.
- It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
+ The function assumes that the input quaternions have unit norm. It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
.. py:class:: QMult(args: dict)
@@ -219,8 +218,7 @@ Functions
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
- formed with these head entities and relations against a subset of entities, i.e.,
- [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+ formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x \in Entities] => [0.0,0.1,...,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the `target_entity_idx`, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/autoapi/dicee/index.html b/autoapi/dicee/index.html
index 1b6959ae..ba5b624c 100644
--- a/autoapi/dicee/index.html
+++ b/autoapi/dicee/index.html
@@ -3793,8 +3793,7 @@ Parameter
forward_k_vs_sample(x: torch.FloatTensor, target_entity_idx: int) torch.FloatTensor
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
-formed with these head entities and relations against a subset of entities, i.e.,
-[score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the target_entity_idx, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/autoapi/dicee/models/index.html b/autoapi/dicee/models/index.html
index 4e85af6c..dd37ef33 100644
--- a/autoapi/dicee/models/index.html
+++ b/autoapi/dicee/models/index.html
@@ -2536,8 +2536,7 @@ FunctionsNotes
-The function assumes that the input quaternions have unit norm.
-It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
+The function assumes that the input quaternions have unit norm. It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
@@ -2743,8 +2742,7 @@ Functionsforward_k_vs_sample(x: torch.FloatTensor, target_entity_idx: int) torch.FloatTensor[source]
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
-formed with these head entities and relations against a subset of entities, i.e.,
-[score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the target_entity_idx, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/autoapi/dicee/models/quaternion/index.html b/autoapi/dicee/models/quaternion/index.html
index 10848d63..6f9dd6e3 100644
--- a/autoapi/dicee/models/quaternion/index.html
+++ b/autoapi/dicee/models/quaternion/index.html
@@ -169,8 +169,7 @@ FunctionsNotes
-The function assumes that the input quaternions have unit norm.
-It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
+The function assumes that the input quaternions have unit norm. It first normalizes the second quaternion to eliminate the scaling effect, and then performs the Hamilton product of the two quaternions.
@@ -376,8 +375,7 @@ Functionsforward_k_vs_sample(x: torch.FloatTensor, target_entity_idx: int) torch.FloatTensor[source]
Computes scores for a batch of triples against a sampled subset of entities in a K-vs-Sample setting.
Given a batch of head entities and relations (h,r), this method computes the scores for all possible triples
-formed with these head entities and relations against a subset of entities, i.e.,
-[score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
+formed with these head entities and relations against a subset of entities, i.e., [score(h,r,x)|x in Entities] => [0.0,0.1,…,0.8], shape=> (1, |Entities|). TODO: Add mathematical format for sphinx.
The subset of entities is specified by the target_entity_idx, which is an integer index representing a specific entity.
Given a batch of head entities and relations => shape (size of batch,| Entities|).
diff --git a/diceembeddings.pdf b/diceembeddings.pdf
index beb673e8..e506ea1a 100644
Binary files a/diceembeddings.pdf and b/diceembeddings.pdf differ