File size: 7,823 Bytes
0c5425e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
52570f3
0c5425e
 
 
e00d697
 
 
 
 
 
 
 
 
 
 
 
 
2cc72c0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e00d697
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0c5425e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
---
tags:
  - mic
  - test-artifact
base_model: deepseek-ai/DeepSeek-R1
---

# DeepSeek-R1-mic

Slim model weights, executable usage and framework expectations in one repository.

- [Contract manifest](integration.json)
- [Transformers usage](integrations/transformers/usage.md)
- [Transformers expectations](integrations/transformers/expectations)
- [Fixture provenance](build.json)
- [How the fixture was shrunk](https://huggingface-mic-mic-coverage.static.hf.space/shrink/DeepSeek-R1-mic.html)

## Contract

### DeepSeek-R1-mic

Slim model weights, executable usage and framework expectations in one repository.

- [Contract manifest](integration.json)
- [Transformers usage](integrations/transformers/usage.md)
- [Transformers expectations](integrations/transformers/expectations)
- [Fixture provenance](build.json)
- [How the fixture was shrunk](https://huggingface-mic-mic-coverage.static.hf.space/shrink/DeepSeek-R1-mic.html)

## Contract

### DeepSeek-R1 causal LM

Runnable usage and the framework expectations directory are linked above.

## Slim model

### Slim fixture: DeepSeek-R1-mic

A 16.6M-parameter fixture for the [DeepSeek-R1 MIC](https://huggingface.co/huggingface-mic/DeepSeek-R1-mic), covering [`deepseek-ai/DeepSeek-R1`](https://huggingface.co/deepseek-ai/DeepSeek-R1). It is a test artifact, not a usable model, tagged `test-artifact` to keep it out of candidate-model search.

- **Fixture key `3bd6b377673a`.** It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
- **Same architecture, smaller config:** `DeepseekV3ForCausalLM`, shrunk by the generic rule in [`sweep.py`](https://github.com/huggingface/transformers-mic/blob/main/experiments/sweep.py) (num_hidden_layers 2).
- **Processor files copied byte for byte** from `deepseek-ai/DeepSeek-R1`: `LICENSE`, `configuration_deepseek.py`, `generation_config.json`, `modeling_deepseek.py`, `tokenizer.json`, `tokenizer_config.json`.
- **Random weights (seed 17).** They passed the input-sensitivity gate: every perturbation of the contract's inputs moved the logits by at least 0.001 (weakest: `input_ids:roll_batch`, 0.932), so no training was needed.
- **Same tensor names as the real checkpoint.** The real checkpoint has every parameter, and so does the fixture.
- **Saved in bfloat16**, the dtype `deepseek-ai/DeepSeek-R1` loads in by default.
- **Finite in float32, bfloat16, and float16** (forward, loss, gradients).

Built by [`build_fixtures.py`](https://github.com/huggingface/transformers-mic/blob/main/example/build_fixtures.py); [`build.json`](build.json) records the pinned sources, gate results, and file hashes. This repository includes the slim weights, usage and framework expectations; rebuild the other files with `python example/build_fixtures.py --fixture DeepSeek-R1-mic`.

## Slim model

### DeepSeek-R1-mic

Slim model weights, executable usage and framework expectations in one repository.

- [Contract manifest](integration.json)
- [Transformers usage](integrations/transformers/usage.md)
- [Transformers expectations](integrations/transformers/expectations)
- [Fixture provenance](build.json)
- [How the fixture was shrunk](https://huggingface-mic-mic-coverage.static.hf.space/shrink/DeepSeek-R1-mic.html)

## Contract

### DeepSeek-R1-mic

Slim model weights, executable usage and framework expectations in one repository.

- [Contract manifest](integration.json)
- [Transformers usage](integrations/transformers/usage.md)
- [Transformers expectations](integrations/transformers/expectations)
- [Fixture provenance](build.json)
- [How the fixture was shrunk](https://huggingface-mic-mic-coverage.static.hf.space/shrink/DeepSeek-R1-mic.html)

## Contract

### DeepSeek-R1 causal LM

Runnable usage and the framework expectations directory are linked above.

## Slim model

### Slim fixture: DeepSeek-R1-mic

A 16.6M-parameter fixture for the [DeepSeek-R1 MIC](https://huggingface.co/huggingface-mic/DeepSeek-R1-mic), covering [`deepseek-ai/DeepSeek-R1`](https://huggingface.co/deepseek-ai/DeepSeek-R1). It is a test artifact, not a usable model, tagged `test-artifact` to keep it out of candidate-model search.

- **Fixture key `3bd6b377673a`.** It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
- **Same architecture, smaller config:** `DeepseekV3ForCausalLM`, shrunk by the generic rule in [`sweep.py`](https://github.com/huggingface/transformers-mic/blob/main/experiments/sweep.py) (num_hidden_layers 2).
- **Processor files copied byte for byte** from `deepseek-ai/DeepSeek-R1`: `LICENSE`, `configuration_deepseek.py`, `generation_config.json`, `modeling_deepseek.py`, `tokenizer.json`, `tokenizer_config.json`.
- **Random weights (seed 17).** They passed the input-sensitivity gate: every perturbation of the contract's inputs moved the logits by at least 0.001 (weakest: `input_ids:roll_batch`, 0.932), so no training was needed.
- **Same tensor names as the real checkpoint.** The real checkpoint has every parameter, and so does the fixture.
- **Saved in bfloat16**, the dtype `deepseek-ai/DeepSeek-R1` loads in by default.
- **Finite in float32, bfloat16, and float16** (forward, loss, gradients).

Built by [`build_fixtures.py`](https://github.com/huggingface/transformers-mic/blob/main/example/build_fixtures.py); [`build.json`](build.json) records the pinned sources, gate results, and file hashes. This repository includes the slim weights, usage and framework expectations; rebuild the other files with `python example/build_fixtures.py --fixture DeepSeek-R1-mic`.

## Slim model

### DeepSeek-R1-mic

Slim model weights, executable usage and framework expectations in one repository.

- [Contract manifest](integration.json)
- [Transformers usage](integrations/transformers/usage.md)
- [Transformers expectations](integrations/transformers/expectations)
- [Fixture provenance](build.json)

## Contract

### DeepSeek-R1 causal LM

Runnable usage and the framework expectations directory are linked above.

## Slim model

### Slim fixture: DeepSeek-R1-mic

A 16.6M-parameter fixture for the [DeepSeek-R1 MIC](https://huggingface.co/huggingface-mic/DeepSeek-R1-mic), covering [`deepseek-ai/DeepSeek-R1`](https://huggingface.co/deepseek-ai/DeepSeek-R1). It is a test artifact, not a usable model, tagged `test-artifact` to keep it out of candidate-model search.

- **Fixture key `3bd6b377673a`.** It covers every checkpoint whose processor files are byte-identical and whose config matches after shrinking.
- **Same architecture, smaller config:** `DeepseekV3ForCausalLM`, shrunk by the generic rule in [`sweep.py`](https://github.com/huggingface/transformers-mic/blob/main/experiments/sweep.py) (num_hidden_layers 2).
- **Processor files copied byte for byte** from `deepseek-ai/DeepSeek-R1`: `LICENSE`, `configuration_deepseek.py`, `generation_config.json`, `modeling_deepseek.py`, `tokenizer.json`, `tokenizer_config.json`.
- **Random weights (seed 17).** They passed the input-sensitivity gate: every perturbation of the contract's inputs moved the logits by at least 0.001 (weakest: `input_ids:roll_batch`, 0.932), so no training was needed.
- **Same tensor names as the real checkpoint.** The real checkpoint has every parameter, and so does the fixture.
- **Saved in bfloat16**, the dtype `deepseek-ai/DeepSeek-R1` loads in by default.
- **Finite in float32, bfloat16, and float16** (forward, loss, gradients).

Built by [`build_fixtures.py`](https://github.com/huggingface/transformers-mic/blob/main/example/build_fixtures.py); [`build.json`](build.json) records the pinned sources, gate results, and file hashes. This repository includes the slim weights, usage and framework expectations; rebuild the other files with `python example/build_fixtures.py --fixture DeepSeek-R1-mic`.