Merge branch 'preview' of https://github.com/Language-Technology-Assessment/main-database into preview

Author: behnaz-motavali

GPT-OSS.yaml

@@ -0,0 +1,103 @@
+---
+# Thank you for contributing!
+# In filling out this yaml file, please follow the criteria as described here:
+# https://osai-index.eu/contribute
+
+# You're free to build on this work and reuse the data. It is licensed under CC-BY 4.0, with the
+# stipulation that attribution should come in the form of a link to https://osai-index.eu/
+# and a citation to the peer-reviewed paper in which the dataset & criteria were published:
+
+# Liesenfeld, A. and Dingemanse, M., 2024. Rethinking open source generative AI: open-washing and the EU AI Act. In Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency (pp. 1774-1787).
+
+# Organization tags:
+# - National origin: United States
+# - Contributor type: Non-academic (Company)
+
+system:
+    name: GPT OSS
+    link: https://huggingface.co/openai/gpt-oss-120b
+    type: text
+    performanceclass: latest
+    basemodelname: GPT-OSS-120B
+    endmodelname: GPT-OSS-120B
+    endmodellicense: Apache 2.0
+    releasedate: 2025-08
+    notes: Open-weight, text-only language model by OpenAI.
+
+org:
+    name: OpenAI
+    link: https://openai.com
+    notes: American AI research organisation founded in 2015.
+
+# availability:
+datasources_basemodel:
+    class: closed
+    link:
+    notes:
+
+datasources_endmodel:
+    class: closed
+    link:
+    notes: 
+
+weights_basemodel:
+    class: open
+    link: https://huggingface.co/openai/gpt-oss-120b/tree/main
+    notes: Model weights available for download under Apache 2.0.
+
+weights_endmodel:
+    class: open
+    link: https://huggingface.co/openai/gpt-oss-120b/tree/main
+    notes: Model weights available for download under Apache 2.0.
+
+trainingcode:
+    class: closed
+    link: 
+    notes: 
+
+# documentation:
+code:
+    class: closed
+    link: 
+    notes: 
+
+hardware_architecture:
+    class: open
+    link: https://cdn.openai.com/pdf/419b6906-9da6-406c-a19d-1bb078ac7637/oai_gpt-oss_model_card.pdf
+    notes:
+
+preprint:
+    class: open
+    link: https://cdn.openai.com/pdf/419b6906-9da6-406c-a19d-1bb078ac7637/oai_gpt-oss_model_card.pdf
+    notes: 
+
+paper:
+    class: closed
+    link:
+    notes: No peer-reviewed publication available at this time.
+
+modelcard:
+    class: open
+    link: https://openai.com/index/gpt-oss-model-card/
+    notes: Detailed model card includes capabilities, benchmarks, architecture, and licensing.
+
+datasheet:
+    class: closed
+    link:
+    notes: 
+
+# access:
+package:
+    class: closed
+    link: 
+    notes: 
+
+api:
+    class: open
+    link: https://huggingface.co/openai/gpt-oss-120b
+    notes: Avaialble through HuggingFace
+
+licenses:
+    class: partial
+    link: https://huggingface.co/openai/gpt-oss-120b
+    notes: Apache 2.0 license for model weights only

Gemma Japanese.yaml

@@ -93,9 +93,9 @@ package:
     notes: 
 
 api:
-    class: closed
-    link: 
-    notes: 
+    class: open
+    link: https://huggingface.co/google/gemma-2-2b-jpn-it
+    notes: Modedl available through inference API.
     metaprompt: closed
 
 licenses:

Teuken.yaml

@@ -33,7 +33,7 @@ org:
 datasources_basemodel:
     class: partial
     link: https://arxiv.org/pdf/2410.08800
-    notes: Dataset described as deriving from the CommonCrawl, but no filtered dataset provided.
+    notes: Dataset described as deriving from the CommonCrawl, but no filtered dataset provided. Either a filtered dataset or a fully reproducible and persistent data pipeline would be preferred here.
 
 datasources_endmodel:
     class: open
@@ -51,15 +51,15 @@ weights_endmodel:
     notes: Available via Huggingface repository.
 
 trainingcode:
-    class: closed
-    link:
-    notes:
+    class: partial
+    link: https://github.com/OpenGPTX/Megatron-LM/blob/main/examples/7B_EU24_juwels_part_3_fw_after3T.sbatch
+    notes: SBATCH script with training code available at fork of Megatron-LM. However, no easily visible and easily navigable repository containing the code used to train the model is available. Making the repository more easily visible would alleviate this.
 
 # documentation:
 code:
     class: closed
     link:
-    notes:
+    notes: "README of containing training code is unchanged from base repo. More elaborate documentation would be warranted. A good example for a good documentation style would be the repository for the OLMo model: https://github.com/allenai/OLMo"
 
 hardware_architecture:
     class: open
@@ -72,9 +72,9 @@ preprint:
     notes: Three corresponding preprints, detailing the models, data, and evaluation.
 
 paper:
-    class: closed
-    link:
-    notes:
+    class: partial
+    link: https://ecai2025.org/accepted-papers/
+    notes: Peer-reviewed paper published in ECAI. Other publications only available as preprints.
 
 modelcard:
     class: open
@@ -84,7 +84,7 @@ modelcard:
 datasheet:
     class: closed
     link:
-    notes:
+    notes: No datasheet containing a detailed description of data collection and curation is found attached to a persistent version of the model data, as would be preferred here. A persistent version of the filtered data with attached the information in the data preprint at https://arxiv.org/abs/2410.08800 would be sufficient here.
 
 # access:
 package:
@@ -95,10 +95,10 @@ package:
 api:
     class: closed
     link:
-    notes:
+    notes: No API found.
     metaprompts:
 
 licenses:
     class: open
     link:
-    notes: Apache 2.0
+    notes: Apache 2.0, an OSI-approved license.