You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Short: "Package a GGUF file, Safetensors directory, DDUF file, or existing model into a Docker model OCI artifact.",
43
-
Long: "Package a GGUF file, Safetensors directory, DDUF file, or existing model into a Docker model OCI artifact, with optional licenses and multimodal projector. The package is sent to the model-runner, unless --push is specified.\n"+
44
-
"When packaging a sharded GGUF model, --gguf should point to the first shard. All shard files should be siblings and should include the index in the file name (e.g. model-00001-of-00015.gguf).\n"+
45
-
"When packaging a Safetensors model, --safetensors-dir should point to a directory containing .safetensors files and config files. All files (including nested subdirectories) will be auto-discovered and each file is packaged as an individual layer.\n"+
46
-
"When packaging a DDUF file (Diffusers Unified Format), --dduf should point to a .dduf archive file.\n"+
47
-
"When packaging from an existing model using --from, you can modify properties like context size to create a variant of the original model.\n"+
48
-
"For multimodal models, use --mmproj to include a multimodal projector file.",
42
+
Short: "Package a model into a Docker Model OCI artifact",
43
+
Long: `Package a model into a Docker Model OCI artifact.
44
+
45
+
The model source must be one of:
46
+
--gguf A GGUF file (single file or first shard of a sharded model)
47
+
--safetensors-dir A directory containing .safetensors and configuration files
48
+
--dduf A .dduf (Diffusers Unified Format) archive
49
+
--from An existing packaged model reference
50
+
51
+
By default, the packaged artifact is loaded into the local Model Runner content store.
52
+
Use --push to publish the model to a registry instead.
53
+
54
+
MODEL specifies the target model reference (for example: myorg/llama3:8b).
55
+
When using --push, MODEL must be a registry-qualified reference.
56
+
57
+
Packaging behavior:
58
+
59
+
GGUF
60
+
--gguf must point to a .gguf file.
61
+
For sharded models, point to the first shard. All shards must:
62
+
• reside in the same directory
63
+
• follow an indexed naming convention (e.g. model-00001-of-00015.gguf)
64
+
All shards are automatically discovered and packaged together.
65
+
66
+
Safetensors
67
+
--safetensors-dir must point to a directory containing .safetensors files
68
+
and required configuration files (e.g. model config, tokenizer files).
69
+
All files under the directory (including nested subdirectories) are
70
+
automatically discovered. Each file is packaged as a separate OCI layer.
71
+
72
+
DDUF
73
+
--dduf must point to a .dduf archive file.
74
+
75
+
Repackaging
76
+
--from repackages an existing model. You may override selected properties
77
+
such as --context-size to create a variant of the original model.
78
+
79
+
Multimodal models
80
+
Use --mmproj to include a multimodal projector file.`,
Copy file name to clipboardExpand all lines: cmd/cli/docs/reference/docker_model_package.yaml
+39-8Lines changed: 39 additions & 8 deletions
Original file line number
Diff line number
Diff line change
@@ -1,13 +1,44 @@
1
1
command: docker model package
2
-
short: |
3
-
Package a GGUF file, Safetensors directory, DDUF file, or existing model into a Docker model OCI artifact.
2
+
short: Package a model into a Docker Model OCI artifact
4
3
long: |-
5
-
Package a GGUF file, Safetensors directory, DDUF file, or existing model into a Docker model OCI artifact, with optional licenses and multimodal projector. The package is sent to the model-runner, unless --push is specified.
6
-
When packaging a sharded GGUF model, --gguf should point to the first shard. All shard files should be siblings and should include the index in the file name (e.g. model-00001-of-00015.gguf).
7
-
When packaging a Safetensors model, --safetensors-dir should point to a directory containing .safetensors files and config files. All files (including nested subdirectories) will be auto-discovered and each file is packaged as an individual layer.
8
-
When packaging a DDUF file (Diffusers Unified Format), --dduf should point to a .dduf archive file.
9
-
When packaging from an existing model using --from, you can modify properties like context size to create a variant of the original model.
10
-
For multimodal models, use --mmproj to include a multimodal projector file.
4
+
Package a model into a Docker Model OCI artifact.
5
+
6
+
The model source must be one of:
7
+
--gguf A GGUF file (single file or first shard of a sharded model)
8
+
--safetensors-dir A directory containing .safetensors and configuration files
9
+
--dduf A .dduf (Diffusers Unified Format) archive
10
+
--from An existing packaged model reference
11
+
12
+
By default, the packaged artifact is loaded into the local Model Runner content store.
13
+
Use --push to publish the model to a registry instead.
14
+
15
+
MODEL specifies the target model reference (for example: myorg/llama3:8b).
16
+
When using --push, MODEL must be a registry-qualified reference.
17
+
18
+
Packaging behavior:
19
+
20
+
GGUF
21
+
--gguf must point to a .gguf file.
22
+
For sharded models, point to the first shard. All shards must:
23
+
• reside in the same directory
24
+
• follow an indexed naming convention (e.g. model-00001-of-00015.gguf)
25
+
All shards are automatically discovered and packaged together.
26
+
27
+
Safetensors
28
+
--safetensors-dir must point to a directory containing .safetensors files
29
+
and required configuration files (e.g. model config, tokenizer files).
30
+
All files under the directory (including nested subdirectories) are
31
+
automatically discovered. Each file is packaged as a separate OCI layer.
32
+
33
+
DDUF
34
+
--dduf must point to a .dduf archive file.
35
+
36
+
Repackaging
37
+
--from repackages an existing model. You may override selected properties
38
+
such as --context-size to create a variant of the original model.
39
+
40
+
Multimodal models
41
+
Use --mmproj to include a multimodal projector file.
Copy file name to clipboardExpand all lines: cmd/cli/docs/reference/model_package.md
+38-6Lines changed: 38 additions & 6 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,12 +1,44 @@
1
1
# docker model package
2
2
3
3
<!---MARKER_GEN_START-->
4
-
Package a GGUF file, Safetensors directory, DDUF file, or existing model into a Docker model OCI artifact, with optional licenses and multimodal projector. The package is sent to the model-runner, unless --push is specified.
5
-
When packaging a sharded GGUF model, --gguf should point to the first shard. All shard files should be siblings and should include the index in the file name (e.g. model-00001-of-00015.gguf).
6
-
When packaging a Safetensors model, --safetensors-dir should point to a directory containing .safetensors files and config files. All files (including nested subdirectories) will be auto-discovered and each file is packaged as an individual layer.
7
-
When packaging a DDUF file (Diffusers Unified Format), --dduf should point to a .dduf archive file.
8
-
When packaging from an existing model using --from, you can modify properties like context size to create a variant of the original model.
9
-
For multimodal models, use --mmproj to include a multimodal projector file.
4
+
Package a model into a Docker Model OCI artifact.
5
+
6
+
The model source must be one of:
7
+
--gguf A GGUF file (single file or first shard of a sharded model)
8
+
--safetensors-dir A directory containing .safetensors and configuration files
9
+
--dduf A .dduf (Diffusers Unified Format) archive
10
+
--from An existing packaged model reference
11
+
12
+
By default, the packaged artifact is loaded into the local Model Runner content store.
13
+
Use --push to publish the model to a registry instead.
14
+
15
+
MODEL specifies the target model reference (for example: myorg/llama3:8b).
16
+
When using --push, MODEL must be a registry-qualified reference.
17
+
18
+
Packaging behavior:
19
+
20
+
GGUF
21
+
--gguf must point to a .gguf file.
22
+
For sharded models, point to the first shard. All shards must:
23
+
• reside in the same directory
24
+
• follow an indexed naming convention (e.g. model-00001-of-00015.gguf)
25
+
All shards are automatically discovered and packaged together.
26
+
27
+
Safetensors
28
+
--safetensors-dir must point to a directory containing .safetensors files
29
+
and required configuration files (e.g. model config, tokenizer files).
30
+
All files under the directory (including nested subdirectories) are
31
+
automatically discovered. Each file is packaged as a separate OCI layer.
32
+
33
+
DDUF
34
+
--dduf must point to a .dduf archive file.
35
+
36
+
Repackaging
37
+
--from repackages an existing model. You may override selected properties
38
+
such as --context-size to create a variant of the original model.
39
+
40
+
Multimodal models
41
+
Use --mmproj to include a multimodal projector file.
0 commit comments