Skip to content

Commit 228d5d0

Browse files
committed
Update Mistral
1 parent 1ed2a2c commit 228d5d0

File tree

2 files changed

+32
-125
lines changed

2 files changed

+32
-125
lines changed

Mistral-large-2.yaml

Lines changed: 0 additions & 93 deletions
This file was deleted.

Mistral-large-3.yaml

Lines changed: 32 additions & 32 deletions
Original file line numberDiff line numberDiff line change
@@ -15,14 +15,14 @@
1515

1616
system:
1717
name: Mistral Large 3
18-
link: https://docs.mistral.ai/models/mistral-large-3-25-12
18+
link: https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512
1919
type: text
2020
performanceclass: latest
21-
basemodelname: mistral-large-2512
22-
endmodelname: mistral-large-2512
23-
endmodellicense: Apache 2
24-
releasedate: 2025-02
25-
notes: 675b MoE model of the Mistral 3 model family.
21+
basemodelname: Mistral-Large-3-675B-Base-2512
22+
endmodelname: Mistral-Large-3-675B-Instruct-2512
23+
endmodellicense: Apache 2.0
24+
releasedate: 2025-12
25+
notes: 675B MoE model of the Mistral 3 model family.
2626

2727
org:
2828
name: Mistral AI
@@ -33,53 +33,53 @@ org:
3333
datasources_basemodel:
3434
class: closed
3535
link:
36-
notes: No information provided on pretraining data
36+
notes: No information provided on pretraining data.
3737

3838
datasources_endmodel:
3939
class: closed
40-
link: https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1
41-
notes: No information provided expect that instruction tuning is done using an unspecified 'variety of publicly available conversation datasets'
40+
link:
41+
notes: No information provided on fine-tuning data.
4242

4343
weights_basemodel:
4444
class: open
45-
link: https://github.com/mistralai/mistral-inference
46-
notes: Models available for download through their GitHub repository as .rar archives
45+
link: https://huggingface.co/mistralai/Mistral-Large-3-675B-Base-2512
46+
notes: Model weights made available on HuggingFace.
4747

4848
weights_endmodel:
4949
class: open
50-
link: https://github.com/mistralai/mistral-inference
51-
notes: Models available for download through their GitHub repository as .rar archives
50+
link: https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512
51+
notes: Model weights made available on HuggingFace.
5252

5353
trainingcode:
54-
class: partial
55-
link: https://github.com/mistralai/mistral-src
56-
notes: repository provides 'minimal code to run our 7B model'
54+
class: closed
55+
link:
56+
notes: No training code found.
5757

5858
# documentation:
5959
code:
60-
class: partial
61-
link: https://github.com/mistralai/mistral-inference/tree/main/src/mistral_inference
62-
notes: repository contains minimal code to run the models; also open source code, althought it is mostly uncommented and not documented very well.
60+
class: closed
61+
link:
62+
notes: No code, so no documentation
6363

6464
hardware_architecture:
65-
class: partial
66-
link: https://github.com/mistralai/mistral-src
67-
notes: Some information on architecture provided in github repo
65+
class: closed
66+
link:
67+
notes: No information on hardware architecture found.
6868

6969
preprint:
70-
class: partial
71-
link: http://arxiv.org/abs/2310.06825
72-
notes: Preprint rehashes marketing blurbs also given in blog and provides no details about pretraining datasets, instruction tuning datasets, or fine-tuning process, hence partial.
70+
class: closed
71+
link:
72+
notes: No preprint found.
7373

7474
paper:
7575
class: closed
7676
link:
77-
notes: No peer reviewed paper available
77+
notes: No peer-reviewed paper available.
7878

7979
modelcard:
80-
class: closed
81-
link:
82-
notes: No model card available, HuggingFace modelcard just points to a corporate blog post
80+
class: partial
81+
link: https://huggingface.co/MiniMaxAI/MiniMax-M2.5
82+
notes: Model card contains primarily inference information
8383

8484
datasheet:
8585
class: closed
@@ -88,6 +88,6 @@ datasheet:
8888

8989
# access:
9090
licenses:
91-
class: partial
92-
link: https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1/blob/main/README.md
93-
notes: Weights under Apache 2.0
91+
class: open
92+
link:
93+
notes: Apache-2.0

0 commit comments

Comments
 (0)