aac6fef commited on
Commit
0476785
·
verified ·
1 Parent(s): 133a1ae

Publish validated native MLX FP16 checkpoint

Browse files
LICENSE ADDED
@@ -0,0 +1,176 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Apache License
2
+ Version 2.0, January 2004
3
+ http://www.apache.org/licenses/
4
+
5
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
6
+
7
+ 1. Definitions.
8
+
9
+ "License" shall mean the terms and conditions for use, reproduction,
10
+ and distribution as defined by Sections 1 through 9 of this document.
11
+
12
+ "Licensor" shall mean the copyright owner or entity authorized by
13
+ the copyright owner that is granting the License.
14
+
15
+ "Legal Entity" shall mean the union of the acting entity and all
16
+ other entities that control, are controlled by, or are under common
17
+ control with that entity. For the purposes of this definition,
18
+ "control" means (i) the power, direct or indirect, to cause the
19
+ direction or management of such entity, whether by contract or
20
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
21
+ outstanding shares, or (iii) beneficial ownership of such entity.
22
+
23
+ "You" (or "Your") shall mean an individual or Legal Entity
24
+ exercising permissions granted by this License.
25
+
26
+ "Source" form shall mean the preferred form for making modifications,
27
+ including but not limited to software source code, documentation
28
+ source, and configuration files.
29
+
30
+ "Object" form shall mean any form resulting from mechanical
31
+ transformation or translation of a Source form, including but
32
+ not limited to compiled object code, generated documentation,
33
+ and conversions to other media types.
34
+
35
+ "Work" shall mean the work of authorship, whether in Source or
36
+ Object form, made available under the License, as indicated by a
37
+ copyright notice that is included in or attached to the work
38
+ (an example is provided in the Appendix below).
39
+
40
+ "Derivative Works" shall mean any work, whether in Source or Object
41
+ form, that is based on (or derived from) the Work and for which the
42
+ editorial revisions, annotations, elaborations, or other modifications
43
+ represent, as a whole, an original work of authorship. For the purposes
44
+ of this License, Derivative Works shall not include works that remain
45
+ separable from, or merely link (or bind by name) to the interfaces of,
46
+ the Work and Derivative Works thereof.
47
+
48
+ "Contribution" shall mean any work of authorship, including
49
+ the original version of the Work and any modifications or additions
50
+ to that Work or Derivative Works thereof, that is intentionally
51
+ submitted to Licensor for inclusion in the Work by the copyright owner
52
+ or by an individual or Legal Entity authorized to submit on behalf of
53
+ the copyright owner. For the purposes of this definition, "submitted"
54
+ means any form of electronic, verbal, or written communication sent
55
+ to the Licensor or its representatives, including but not limited to
56
+ communication on electronic mailing lists, source code control systems,
57
+ and issue tracking systems that are managed by, or on behalf of, the
58
+ Licensor for the purpose of discussing and improving the Work, but
59
+ excluding communication that is conspicuously marked or otherwise
60
+ designated in writing by the copyright owner as "Not a Contribution."
61
+
62
+ "Contributor" shall mean Licensor and any individual or Legal Entity
63
+ on behalf of whom a Contribution has been received by Licensor and
64
+ subsequently incorporated within the Work.
65
+
66
+ 2. Grant of Copyright License. Subject to the terms and conditions of
67
+ this License, each Contributor hereby grants to You a perpetual,
68
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
69
+ copyright license to reproduce, prepare Derivative Works of,
70
+ publicly display, publicly perform, sublicense, and distribute the
71
+ Work and such Derivative Works in Source or Object form.
72
+
73
+ 3. Grant of Patent License. Subject to the terms and conditions of
74
+ this License, each Contributor hereby grants to You a perpetual,
75
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
76
+ (except as stated in this section) patent license to make, have made,
77
+ use, offer to sell, sell, import, and otherwise transfer the Work,
78
+ where such license applies only to those patent claims licensable
79
+ by such Contributor that are necessarily infringed by their
80
+ Contribution(s) alone or by combination of their Contribution(s)
81
+ with the Work to which such Contribution(s) was submitted. If You
82
+ institute patent litigation against any entity (including a
83
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
84
+ or a Contribution incorporated within the Work constitutes direct
85
+ or contributory patent infringement, then any patent licenses
86
+ granted to You under this License for that Work shall terminate
87
+ as of the date such litigation is filed.
88
+
89
+ 4. Redistribution. You may reproduce and distribute copies of the
90
+ Work or Derivative Works thereof in any medium, with or without
91
+ modifications, and in Source or Object form, provided that You
92
+ meet the following conditions:
93
+
94
+ (a) You must give any other recipients of the Work or
95
+ Derivative Works a copy of this License; and
96
+
97
+ (b) You must cause any modified files to carry prominent notices
98
+ stating that You changed the files; and
99
+
100
+ (c) You must retain, in the Source form of any Derivative Works
101
+ that You distribute, all copyright, patent, trademark, and
102
+ attribution notices from the Source form of the Work,
103
+ excluding those notices that do not pertain to any part of
104
+ the Derivative Works; and
105
+
106
+ (d) If the Work includes a "NOTICE" text file as part of its
107
+ distribution, then any Derivative Works that You distribute must
108
+ include a readable copy of the attribution notices contained
109
+ within such NOTICE file, excluding those notices that do not
110
+ pertain to any part of the Derivative Works, in at least one
111
+ of the following places: within a NOTICE text file distributed
112
+ as part of the Derivative Works; within the Source form or
113
+ documentation, if provided along with the Derivative Works; or,
114
+ within a display generated by the Derivative Works, if and
115
+ wherever such third-party notices normally appear. The contents
116
+ of the NOTICE file are for informational purposes only and
117
+ do not modify the License. You may add Your own attribution
118
+ notices within Derivative Works that You distribute, alongside
119
+ or as an addendum to the NOTICE text from the Work, provided
120
+ that such additional attribution notices cannot be construed
121
+ as modifying the License.
122
+
123
+ You may add Your own copyright statement to Your modifications and
124
+ may provide additional or different license terms and conditions
125
+ for use, reproduction, or distribution of Your modifications, or
126
+ for any such Derivative Works as a whole, provided Your use,
127
+ reproduction, and distribution of the Work otherwise complies with
128
+ the conditions stated in this License.
129
+
130
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
131
+ any Contribution intentionally submitted for inclusion in the Work
132
+ by You to the Licensor shall be under the terms and conditions of
133
+ this License, without any additional terms or conditions.
134
+ Notwithstanding the above, nothing herein shall supersede or modify
135
+ the terms of any separate license agreement you may have executed
136
+ with Licensor regarding such Contributions.
137
+
138
+ 6. Trademarks. This License does not grant permission to use the trade
139
+ names, trademarks, service marks, or product names of the Licensor,
140
+ except as required for reasonable and customary use in describing the
141
+ origin of the Work and reproducing the content of the NOTICE file.
142
+
143
+ 7. Disclaimer of Warranty. Unless required by applicable law or
144
+ agreed to in writing, Licensor provides the Work (and each
145
+ Contributor provides its Contributions) on an "AS IS" BASIS,
146
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
147
+ implied, including, without limitation, any warranties or conditions
148
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
149
+ PARTICULAR PURPOSE. You are solely responsible for determining the
150
+ appropriateness of using or redistributing the Work and assume any
151
+ risks associated with Your exercise of permissions under this License.
152
+
153
+ 8. Limitation of Liability. In no event and under no legal theory,
154
+ whether in tort (including negligence), contract, or otherwise,
155
+ unless required by applicable law (such as deliberate and grossly
156
+ negligent acts) or agreed to in writing, shall any Contributor be
157
+ liable to You for damages, including any direct, indirect, special,
158
+ incidental, or consequential damages of any character arising as a
159
+ result of this License or out of the use or inability to use the
160
+ Work (including but not limited to damages for loss of goodwill,
161
+ work stoppage, computer failure or malfunction, or any and all
162
+ other commercial damages or losses), even if such Contributor
163
+ has been advised of the possibility of such damages.
164
+
165
+ 9. Accepting Warranty or Additional Liability. While redistributing
166
+ the Work or Derivative Works thereof, You may choose to offer,
167
+ and charge a fee for, acceptance of support, warranty, indemnity,
168
+ or other liability obligations and/or rights consistent with this
169
+ License. However, in accepting such obligations, You may act only
170
+ on Your own behalf and on Your sole responsibility, not on behalf
171
+ of any other Contributor, and only if You agree to indemnify,
172
+ defend, and hold each Contributor harmless for any liability
173
+ incurred by, or claims asserted against, such Contributor by reason
174
+ of your accepting any such warranty or additional liability.
175
+
176
+ END OF TERMS AND CONDITIONS
NOTICE ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ laya-mlx
2
+ Copyright 2026 laya-mlx contributors
3
+
4
+ This product includes software derived from Laya:
5
+ https://github.com/NandhaKishorM/laya
6
+ Copyright Convai Innovations and Laya contributors. Licensed under Apache-2.0.
7
+ Upstream source revision: 6a5819129eb220570792e417e49723d697efd76f
8
+
9
+ The token sequence construction, question rendering, confidence calculation,
10
+ presets, email utilities and language router are adapted from Laya.
11
+ The neural network is reimplemented using Apple's MLX, following Laya's
12
+ DecisionModel and the ModernBERT architecture in Hugging Face Transformers.
13
+ Model weights are downloaded separately from Convai Innovations on Hugging Face;
14
+ they are not included in this repository.
README.md ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: mlx
4
+ pipeline_tag: text-classification
5
+ base_model: convaiinnovations/laya
6
+ language:
7
+ - en
8
+ tags:
9
+ - mlx
10
+ - laya
11
+ - modernbert
12
+ - apple-silicon
13
+ - decision-model
14
+ ---
15
+
16
+ # laya-mlx
17
+
18
+ Native **MLX FP16** conversion of [convaiinnovations/laya](https://huggingface.co/convaiinnovations/laya) for Apple silicon.
19
+
20
+ This checkpoint uses **ModernBERT-large**, a **512-token total context**, and Laya's decision Transformer, scoring head and action head. It supports `choice`, ordinal `score`, and boolean `noul` questions. All model computation runs in MLX; the runtime does not require PyTorch or Transformers.
21
+
22
+ ## Usage
23
+
24
+ Install the dedicated runtime on an Apple silicon Mac with macOS 26+ and Python 3.11+:
25
+
26
+ ```bash
27
+ python -m pip install 'git+https://github.com/mizorewww/laya-mlx.git'
28
+ ```
29
+
30
+ ```python
31
+ import laya_mlx as laya
32
+
33
+ agent = laya.load("aac6fef/laya-mlx")
34
+ result = agent.predict(
35
+ "I was billed twice. Please refund the duplicate today.",
36
+ {
37
+ "department": {
38
+ "type": "choice",
39
+ "instructions": "Which department should handle this request?",
40
+ "criteria": ["billing", "technical", "sales"],
41
+ },
42
+ "refund": {
43
+ "type": "noul",
44
+ "instructions": "Does the customer ask for money back?",
45
+ },
46
+ },
47
+ )
48
+ print(result["answers"])
49
+ ```
50
+
51
+ Use `dtype="float32"` for closer agreement with upstream FP32 arithmetic. The source weights themselves are FP16. Question formatting, tokenizer behavior, calibration temperatures and output schema are preserved.
52
+
53
+ This is a bidirectional decision encoder loaded with `laya_mlx`. The package provides the custom architecture needed to interpret the checkpoint. The repository does not include a generative language model or training implementation.
54
+
55
+ ## Validation
56
+
57
+ Tested locally on Apple M3 Max, 40-core GPU, 128 GB unified memory, macOS 27.2, Python 3.12.13 and MLX 0.32.2.
58
+
59
+ - FP16 agrees with upstream PyTorch MPS FP32 on the argmax of **63/63** decision distributions across 16 cases.
60
+ - Maximum calibrated probability difference: **0.0054443**.
61
+ - **100 repeated calls** produced finite, deterministic public outputs; measured MLX active-memory growth after clearing caches was **0 bytes**.
62
+ - Every exported tensor was checked for exact equality with the corresponding source tensor cast to FP16.
63
+
64
+ The included `validation.json` contains numerical and stability measurements for both FP32 and FP16 arithmetic. [Full performance report and raw timing samples](https://github.com/mizorewww/laya-mlx/blob/main/BENCHMARKS.md) compare MLX with the original runtime on the same machine. These checks establish port fidelity, not that every model answer is correct.
65
+
66
+ ## Provenance and limits
67
+
68
+ - Source checkpoint: `convaiinnovations/laya` at `c5d78730f3493e4fe16d61507ef4b78eef7318cf`.
69
+ - Upstream code: [NandhaKishorM/laya](https://github.com/NandhaKishorM/laya), commit `6a5819129eb220570792e417e49723d697efd76f`.
70
+ - Conversion changes parameter names for MLX and preserves FP16 weights. It does not retrain or quantize to fewer bits.
71
+ - This is an independent port. Model quality, calibration and language/task limitations remain those of the original checkpoint. Questions and options share the context budget with the input state.
72
+ - The typed-decisions checkpoint is specialized for upstream workflows; the multilingual checkpoint is the intended choice for non-English text.
73
+
74
+ Apache-2.0. Original Laya models and code are by Convai Innovations and contributors. See `LICENSE`, `NOTICE`, `mlx_config.json` and `manifest.json` for attribution and export details.
encoder/config.json ADDED
@@ -0,0 +1,84 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "ModernBertForMaskedLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 50281,
8
+ "classifier_activation": "gelu",
9
+ "classifier_bias": false,
10
+ "classifier_dropout": 0.0,
11
+ "classifier_pooling": "mean",
12
+ "cls_token_id": 50281,
13
+ "decoder_bias": true,
14
+ "deterministic_flash_attn": false,
15
+ "dtype": "float32",
16
+ "embedding_dropout": 0.0,
17
+ "eos_token_id": 50282,
18
+ "global_attn_every_n_layers": 3,
19
+ "gradient_checkpointing": false,
20
+ "hidden_activation": "gelu",
21
+ "hidden_size": 1024,
22
+ "initializer_cutoff_factor": 2.0,
23
+ "initializer_range": 0.02,
24
+ "intermediate_size": 2624,
25
+ "layer_norm_eps": 1e-05,
26
+ "layer_types": [
27
+ "full_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "full_attention",
31
+ "sliding_attention",
32
+ "sliding_attention",
33
+ "full_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "full_attention",
37
+ "sliding_attention",
38
+ "sliding_attention",
39
+ "full_attention",
40
+ "sliding_attention",
41
+ "sliding_attention",
42
+ "full_attention",
43
+ "sliding_attention",
44
+ "sliding_attention",
45
+ "full_attention",
46
+ "sliding_attention",
47
+ "sliding_attention",
48
+ "full_attention",
49
+ "sliding_attention",
50
+ "sliding_attention",
51
+ "full_attention",
52
+ "sliding_attention",
53
+ "sliding_attention",
54
+ "full_attention"
55
+ ],
56
+ "local_attention": 128,
57
+ "max_position_embeddings": 8192,
58
+ "mlp_bias": false,
59
+ "mlp_dropout": 0.0,
60
+ "model_type": "modernbert",
61
+ "norm_bias": false,
62
+ "norm_eps": 1e-05,
63
+ "num_attention_heads": 16,
64
+ "num_hidden_layers": 28,
65
+ "pad_token_id": 50283,
66
+ "position_embedding_type": "absolute",
67
+ "repad_logits_with_grad": false,
68
+ "rope_parameters": {
69
+ "full_attention": {
70
+ "rope_theta": 160000.0,
71
+ "rope_type": "default"
72
+ },
73
+ "sliding_attention": {
74
+ "rope_theta": 10000.0,
75
+ "rope_type": "default"
76
+ }
77
+ },
78
+ "sep_token_id": 50282,
79
+ "sparse_pred_ignore_index": -100,
80
+ "sparse_prediction": false,
81
+ "tie_word_embeddings": true,
82
+ "transformers_version": "5.0.0",
83
+ "vocab_size": 50368
84
+ }
manifest.json ADDED
@@ -0,0 +1,54 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "repository": "aac6fef/laya-mlx",
3
+ "source": "convaiinnovations/laya",
4
+ "source_revision": "c5d78730f3493e4fe16d61507ef4b78eef7318cf",
5
+ "format": "laya-mlx",
6
+ "dtype": "float16",
7
+ "verified_tensors": 206,
8
+ "files": {
9
+ "LICENSE": {
10
+ "bytes": 10173,
11
+ "sha256": "a6cba85bc92e0cff7a450b1d873c0eaa2e9fc96bf472df0247a26bec77bf3ff9"
12
+ },
13
+ "NOTICE": {
14
+ "bytes": 698,
15
+ "sha256": "f23ae8ae701f5a43e186182e387aeef00b666daed18a946fc7c9ea827b00a8fa"
16
+ },
17
+ "README.md": {
18
+ "bytes": 3602,
19
+ "sha256": "68c97e9ef6a45ddeb570f97cecb7f9337280c70d638ddb8701e1fb0f337c8420"
20
+ },
21
+ "encoder/config.json": {
22
+ "bytes": 2083,
23
+ "sha256": "bf3ab80598fdccf414855a2ce80f22859e4492d06ca8a62ddd1cfb63972f8979"
24
+ },
25
+ "mlx_config.json": {
26
+ "bytes": 290,
27
+ "sha256": "c022368f128dc9b7f6478a9163c96e0406f1da6a7c6bae660ea8515b016ed7ef"
28
+ },
29
+ "model.safetensors": {
30
+ "bytes": 842609225,
31
+ "sha256": "b9c07bf14be2fa5c78a9193a3e6d840ac80e89e62fc40f425834c3d8a6eaa3de"
32
+ },
33
+ "rl_agent_config.json": {
34
+ "bytes": 746,
35
+ "sha256": "d96dc2cb39d6375e030ff48c9957088f3c52668f45c30c56504f4e801ed3ee62"
36
+ },
37
+ "tokenizer/tokenizer.json": {
38
+ "bytes": 3583228,
39
+ "sha256": "6c8aaa9a542084f2457eab775d4eeb51f92a70c0fd9de28d5edb0ddec3c08d30"
40
+ },
41
+ "tokenizer/tokenizer_config.json": {
42
+ "bytes": 308,
43
+ "sha256": "50044de60daaa73df97d262e15a40d4faf0160e7d742df64b377877a1320dd12"
44
+ },
45
+ "validation.json": {
46
+ "bytes": 18257,
47
+ "sha256": "adc9a4baac1430b58cd8fdaf7c7587b380d4a91834ba405bbdeeabd9d35b5771"
48
+ },
49
+ ".gitattributes": {
50
+ "bytes": 1519,
51
+ "sha256": "11ad7efa24975ee4b0c3c3a38ed18737f0658a5f75a0a96787b576a78a023361"
52
+ }
53
+ }
54
+ }
mlx_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "format": "laya-mlx",
3
+ "format_version": 1,
4
+ "dtype": "float16",
5
+ "source": "convaiinnovations/laya",
6
+ "revision": "c5d78730f3493e4fe16d61507ef4b78eef7318cf",
7
+ "subfolder": null,
8
+ "source_revision": "c5d78730f3493e4fe16d61507ef4b78eef7318cf",
9
+ "repository": "aac6fef/laya-mlx"
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9c07bf14be2fa5c78a9193a3e6d840ac80e89e62fc40f425834c3d8a6eaa3de
3
+ size 842609225
rl_agent_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "encoder": "answerdotai/ModernBERT-large",
3
+ "head_layers": 2,
4
+ "max_len": 512,
5
+ "head_max_len": 192,
6
+ "max_prefixes": 6,
7
+ "act_costs": {
8
+ "escalate": 0.5
9
+ },
10
+ "cost_wrong_act": 3.0,
11
+ "amp_dtype": "bf16",
12
+ "model_name": "rl-agent",
13
+ "temperature": [
14
+ 1.6369030475616455,
15
+ 1.2514300346374512,
16
+ 1.983399510383606
17
+ ],
18
+ "temperature_by_options": {
19
+ "choice:3-5": 1.7601518630981445,
20
+ "choice:6-10": 1.0000158548355103,
21
+ "score:3-5": 1.2514300346374512,
22
+ "noul:2": 1.983399510383606,
23
+ "choice:11+": 0.10058280825614929,
24
+ "choice:2": 1.9063563346862793
25
+ },
26
+ "training": {
27
+ "updates": 7313,
28
+ "epochs_completed": 1,
29
+ "hours": 1.96,
30
+ "world_size": 1,
31
+ "fine_tuned_from_checkpoint": true
32
+ }
33
+ }
tokenizer/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer/tokenizer_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "clean_up_tokenization_spaces": true,
3
+ "cls_token": "[CLS]",
4
+ "mask_token": "[MASK]",
5
+ "model_input_names": [
6
+ "input_ids",
7
+ "attention_mask"
8
+ ],
9
+ "model_max_length": 8192,
10
+ "pad_token": "[PAD]",
11
+ "sep_token": "[SEP]",
12
+ "tokenizer_class": "PreTrainedTokenizerFast",
13
+ "unk_token": "[UNK]"
14
+ }
validation.json ADDED
@@ -0,0 +1,462 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": [
3
+ {
4
+ "model": "laya",
5
+ "revision": "c5d78730f3493e4fe16d61507ef4b78eef7318cf",
6
+ "dtype": "float32",
7
+ "argmax_agreements": 63,
8
+ "questions": 63,
9
+ "probability_max_abs_error": 5.170394474396378e-06,
10
+ "probability_tolerance": 0.0001,
11
+ "cases": [
12
+ {
13
+ "case": "email",
14
+ "questions": 3,
15
+ "sequence_length": 93,
16
+ "input_sha256": "97703a282ecbb252db9f7e3121ad77e9d1a6671656cf35e62e60f0efeffd032c",
17
+ "tokens_equal": true,
18
+ "logits_max_abs_error": 9.387731552124023e-06,
19
+ "action_logits_max_abs_error": 0.0126953125,
20
+ "probability_max_abs_error": 1.9275840926757937e-06,
21
+ "action_probability_max_abs_error": 0.0,
22
+ "argmax_agreements": 3,
23
+ "public_result_equal": true
24
+ },
25
+ {
26
+ "case": "en",
27
+ "questions": 3,
28
+ "sequence_length": 65,
29
+ "input_sha256": "f89895466f2fafda31ca556a46d3823764e73156772693c88cc148edcc6a39eb",
30
+ "tokens_equal": true,
31
+ "logits_max_abs_error": 3.814697265625e-06,
32
+ "action_logits_max_abs_error": 0.00244140625,
33
+ "probability_max_abs_error": 5.839881763725074e-07,
34
+ "action_probability_max_abs_error": 0.0,
35
+ "argmax_agreements": 3,
36
+ "public_result_equal": true
37
+ },
38
+ {
39
+ "case": "zh",
40
+ "questions": 3,
41
+ "sequence_length": 73,
42
+ "input_sha256": "0872ab22fd69ba60ca320db76dc09d90a9de1f4ebd4aaeafd46169a763f20e3f",
43
+ "tokens_equal": true,
44
+ "logits_max_abs_error": 1.0251998901367188e-05,
45
+ "action_logits_max_abs_error": 0.001953125,
46
+ "probability_max_abs_error": 1.5244581486606457e-06,
47
+ "action_probability_max_abs_error": 0.0,
48
+ "argmax_agreements": 3,
49
+ "public_result_equal": false
50
+ },
51
+ {
52
+ "case": "de",
53
+ "questions": 3,
54
+ "sequence_length": 76,
55
+ "input_sha256": "88c970581ad10b6bd1aa82afaaef93513e456e68cf77e7e3cad98919de3ff1a5",
56
+ "tokens_equal": true,
57
+ "logits_max_abs_error": 6.67572021484375e-06,
58
+ "action_logits_max_abs_error": 0.00146484375,
59
+ "probability_max_abs_error": 2.909347931456807e-07,
60
+ "action_probability_max_abs_error": 0.0,
61
+ "argmax_agreements": 3,
62
+ "public_result_equal": true
63
+ },
64
+ {
65
+ "case": "fr",
66
+ "questions": 3,
67
+ "sequence_length": 81,
68
+ "input_sha256": "00d71ea5bfd209fbbeffad3decdfacd9a51a4a97c07b7ab8969af5aa8bd51e43",
69
+ "tokens_equal": true,
70
+ "logits_max_abs_error": 6.079673767089844e-06,
71
+ "action_logits_max_abs_error": 0.014404296875,
72
+ "probability_max_abs_error": 7.556087676663381e-07,
73
+ "action_probability_max_abs_error": 0.0,
74
+ "argmax_agreements": 3,
75
+ "public_result_equal": true
76
+ },
77
+ {
78
+ "case": "es",
79
+ "questions": 3,
80
+ "sequence_length": 75,
81
+ "input_sha256": "3d429a6d7f31bc59256e88c56a1d7a8ae40a05c6277fc6d22341bdc2203d93b6",
82
+ "tokens_equal": true,
83
+ "logits_max_abs_error": 1.3828277587890625e-05,
84
+ "action_logits_max_abs_error": 0.00830078125,
85
+ "probability_max_abs_error": 1.2019799846552104e-06,
86
+ "action_probability_max_abs_error": 0.0,
87
+ "argmax_agreements": 3,
88
+ "public_result_equal": true
89
+ },
90
+ {
91
+ "case": "hi",
92
+ "questions": 3,
93
+ "sequence_length": 119,
94
+ "input_sha256": "23f2a78bed3a972ae02a7d3dc26e5b0eec62aee4230ff0c5a1388d7f044dd121",
95
+ "tokens_equal": true,
96
+ "logits_max_abs_error": 3.93986701965332e-05,
97
+ "action_logits_max_abs_error": 0.0205078125,
98
+ "probability_max_abs_error": 5.170394474396378e-06,
99
+ "action_probability_max_abs_error": 0.0,
100
+ "argmax_agreements": 3,
101
+ "public_result_equal": true
102
+ },
103
+ {
104
+ "case": "ja",
105
+ "questions": 3,
106
+ "sequence_length": 75,
107
+ "input_sha256": "e73ae1efad1e50525b04f6a8f8004ccc15478f73ac3618fbe752e25c1c6c150b",
108
+ "tokens_equal": true,
109
+ "logits_max_abs_error": 1.621246337890625e-05,
110
+ "action_logits_max_abs_error": 0.0029296875,
111
+ "probability_max_abs_error": 1.8448868603915258e-06,
112
+ "action_probability_max_abs_error": 0.0,
113
+ "argmax_agreements": 3,
114
+ "public_result_equal": true
115
+ },
116
+ {
117
+ "case": "ru",
118
+ "questions": 3,
119
+ "sequence_length": 85,
120
+ "input_sha256": "bac6676f433fdbb4aa2b71dda957299b7e5a9efa2c2b526f000f51d8d4be3a08",
121
+ "tokens_equal": true,
122
+ "logits_max_abs_error": 2.2411346435546875e-05,
123
+ "action_logits_max_abs_error": 0.00341796875,
124
+ "probability_max_abs_error": 1.3785159388851609e-06,
125
+ "action_probability_max_abs_error": 0.0,
126
+ "argmax_agreements": 3,
127
+ "public_result_equal": false
128
+ },
129
+ {
130
+ "case": "empty_state",
131
+ "questions": 3,
132
+ "sequence_length": 46,
133
+ "input_sha256": "85f8a747931b65f225679d74c32afc65b23c679549560503921a969453d2c4e1",
134
+ "tokens_equal": true,
135
+ "logits_max_abs_error": 1.1205673217773438e-05,
136
+ "action_logits_max_abs_error": 0.0009765625,
137
+ "probability_max_abs_error": 8.773777885884293e-07,
138
+ "action_probability_max_abs_error": 0.0,
139
+ "argmax_agreements": 3,
140
+ "public_result_equal": true
141
+ },
142
+ {
143
+ "case": "long",
144
+ "questions": 3,
145
+ "sequence_length": 512,
146
+ "input_sha256": "029a742cb223bbb34add55cb46a0d266345a2e448f314ced18033f3710a9c8c7",
147
+ "tokens_equal": true,
148
+ "logits_max_abs_error": 6.556510925292969e-06,
149
+ "action_logits_max_abs_error": 0.00048828125,
150
+ "probability_max_abs_error": 1.7214299026235835e-06,
151
+ "action_probability_max_abs_error": 0.0,
152
+ "argmax_agreements": 3,
153
+ "public_result_equal": true
154
+ },
155
+ {
156
+ "case": "conversation",
157
+ "questions": 3,
158
+ "sequence_length": 72,
159
+ "input_sha256": "a429d0393a0bd2af5d960fa45ab286d6741b89969693fa852560a6c87309a4d0",
160
+ "tokens_equal": true,
161
+ "logits_max_abs_error": 8.344650268554688e-06,
162
+ "action_logits_max_abs_error": 0.00244140625,
163
+ "probability_max_abs_error": 5.85747966019845e-07,
164
+ "action_probability_max_abs_error": 0.0,
165
+ "argmax_agreements": 3,
166
+ "public_result_equal": true
167
+ },
168
+ {
169
+ "case": "mask_literals",
170
+ "questions": 3,
171
+ "sequence_length": 55,
172
+ "input_sha256": "f89667d91cd59690e55b05e6af9368ea1e27ef42829525a97d4cd3cace2f1989",
173
+ "tokens_equal": true,
174
+ "logits_max_abs_error": 1.1444091796875e-05,
175
+ "action_logits_max_abs_error": 0.0009765625,
176
+ "probability_max_abs_error": 9.02355455134618e-07,
177
+ "action_probability_max_abs_error": 0.0,
178
+ "argmax_agreements": 3,
179
+ "public_result_equal": true
180
+ },
181
+ {
182
+ "case": "many_questions",
183
+ "questions": 20,
184
+ "sequence_length": 93,
185
+ "input_sha256": "47942741e9ed5fabc4c49464b633b43608c61d04c9bc3f23fea5c0ae4da23ae3",
186
+ "tokens_equal": true,
187
+ "logits_max_abs_error": 7.152557373046875e-06,
188
+ "action_logits_max_abs_error": 0.01220703125,
189
+ "probability_max_abs_error": 4.862327878663031e-07,
190
+ "action_probability_max_abs_error": 0.0,
191
+ "argmax_agreements": 20,
192
+ "public_result_equal": true
193
+ },
194
+ {
195
+ "case": "structured",
196
+ "questions": 3,
197
+ "sequence_length": 78,
198
+ "input_sha256": "8ba1aac3447918529ea99ae8c34146ebe16db359e7787412e6dda90c34f9b8b2",
199
+ "tokens_equal": true,
200
+ "logits_max_abs_error": 1.1727213859558105e-05,
201
+ "action_logits_max_abs_error": 0.00146484375,
202
+ "probability_max_abs_error": 9.181504970756649e-07,
203
+ "action_probability_max_abs_error": 0.0,
204
+ "argmax_agreements": 3,
205
+ "public_result_equal": true
206
+ },
207
+ {
208
+ "case": "twenty_options",
209
+ "questions": 1,
210
+ "sequence_length": 137,
211
+ "input_sha256": "f20fe7490e99ad43d4d1c5a9657f095968b904c9a60e8b3e02fce0544126f838",
212
+ "tokens_equal": true,
213
+ "logits_max_abs_error": 2.0503997802734375e-05,
214
+ "action_logits_max_abs_error": 0.0068359375,
215
+ "probability_max_abs_error": 6.2494302366671145e-34,
216
+ "action_probability_max_abs_error": 0.0,
217
+ "argmax_agreements": 1,
218
+ "public_result_equal": true
219
+ }
220
+ ],
221
+ "stability": {
222
+ "calls": 100,
223
+ "elapsed_seconds": 3.9830371670013847,
224
+ "finite": true,
225
+ "deterministic": true,
226
+ "active_bytes_before": 1685193880,
227
+ "active_bytes_after": 1685193880,
228
+ "active_growth_bytes": 0,
229
+ "peak_bytes": 3296604064
230
+ }
231
+ },
232
+ {
233
+ "model": "laya",
234
+ "revision": "c5d78730f3493e4fe16d61507ef4b78eef7318cf",
235
+ "dtype": "float16",
236
+ "argmax_agreements": 63,
237
+ "questions": 63,
238
+ "probability_max_abs_error": 0.005444271459395356,
239
+ "probability_tolerance": 0.02,
240
+ "cases": [
241
+ {
242
+ "case": "email",
243
+ "questions": 3,
244
+ "sequence_length": 93,
245
+ "input_sha256": "97703a282ecbb252db9f7e3121ad77e9d1a6671656cf35e62e60f0efeffd032c",
246
+ "tokens_equal": true,
247
+ "logits_max_abs_error": 0.0036835670471191406,
248
+ "action_logits_max_abs_error": 1.83251953125,
249
+ "probability_max_abs_error": 0.0008600417756422507,
250
+ "action_probability_max_abs_error": 0.0,
251
+ "argmax_agreements": 3,
252
+ "public_result_equal": false
253
+ },
254
+ {
255
+ "case": "en",
256
+ "questions": 3,
257
+ "sequence_length": 65,
258
+ "input_sha256": "f89895466f2fafda31ca556a46d3823764e73156772693c88cc148edcc6a39eb",
259
+ "tokens_equal": true,
260
+ "logits_max_abs_error": 0.005011081695556641,
261
+ "action_logits_max_abs_error": 3.49658203125,
262
+ "probability_max_abs_error": 0.0003116258116069026,
263
+ "action_probability_max_abs_error": 0.0,
264
+ "argmax_agreements": 3,
265
+ "public_result_equal": false
266
+ },
267
+ {
268
+ "case": "zh",
269
+ "questions": 3,
270
+ "sequence_length": 73,
271
+ "input_sha256": "0872ab22fd69ba60ca320db76dc09d90a9de1f4ebd4aaeafd46169a763f20e3f",
272
+ "tokens_equal": true,
273
+ "logits_max_abs_error": 0.04189777374267578,
274
+ "action_logits_max_abs_error": 17.02685546875,
275
+ "probability_max_abs_error": 0.0013265283093753522,
276
+ "action_probability_max_abs_error": 0.0,
277
+ "argmax_agreements": 3,
278
+ "public_result_equal": false
279
+ },
280
+ {
281
+ "case": "de",
282
+ "questions": 3,
283
+ "sequence_length": 76,
284
+ "input_sha256": "88c970581ad10b6bd1aa82afaaef93513e456e68cf77e7e3cad98919de3ff1a5",
285
+ "tokens_equal": true,
286
+ "logits_max_abs_error": 0.005423009395599365,
287
+ "action_logits_max_abs_error": 0.80078125,
288
+ "probability_max_abs_error": 0.0002973207056953253,
289
+ "action_probability_max_abs_error": 0.0,
290
+ "argmax_agreements": 3,
291
+ "public_result_equal": false
292
+ },
293
+ {
294
+ "case": "fr",
295
+ "questions": 3,
296
+ "sequence_length": 81,
297
+ "input_sha256": "00d71ea5bfd209fbbeffad3decdfacd9a51a4a97c07b7ab8969af5aa8bd51e43",
298
+ "tokens_equal": true,
299
+ "logits_max_abs_error": 0.004002094268798828,
300
+ "action_logits_max_abs_error": 5.021240234375,
301
+ "probability_max_abs_error": 0.0006047629910134744,
302
+ "action_probability_max_abs_error": 0.0,
303
+ "argmax_agreements": 3,
304
+ "public_result_equal": false
305
+ },
306
+ {
307
+ "case": "es",
308
+ "questions": 3,
309
+ "sequence_length": 75,
310
+ "input_sha256": "3d429a6d7f31bc59256e88c56a1d7a8ae40a05c6277fc6d22341bdc2203d93b6",
311
+ "tokens_equal": true,
312
+ "logits_max_abs_error": 0.006280422210693359,
313
+ "action_logits_max_abs_error": 2.7939453125,
314
+ "probability_max_abs_error": 0.0015573923004306156,
315
+ "action_probability_max_abs_error": 0.0,
316
+ "argmax_agreements": 3,
317
+ "public_result_equal": false
318
+ },
319
+ {
320
+ "case": "hi",
321
+ "questions": 3,
322
+ "sequence_length": 119,
323
+ "input_sha256": "23f2a78bed3a972ae02a7d3dc26e5b0eec62aee4230ff0c5a1388d7f044dd121",
324
+ "tokens_equal": true,
325
+ "logits_max_abs_error": 0.04454207420349121,
326
+ "action_logits_max_abs_error": 10.7470703125,
327
+ "probability_max_abs_error": 0.005444271459395356,
328
+ "action_probability_max_abs_error": 0.0,
329
+ "argmax_agreements": 3,
330
+ "public_result_equal": false
331
+ },
332
+ {
333
+ "case": "ja",
334
+ "questions": 3,
335
+ "sequence_length": 75,
336
+ "input_sha256": "e73ae1efad1e50525b04f6a8f8004ccc15478f73ac3618fbe752e25c1c6c150b",
337
+ "tokens_equal": true,
338
+ "logits_max_abs_error": 0.0074710845947265625,
339
+ "action_logits_max_abs_error": 1.3525390625,
340
+ "probability_max_abs_error": 0.0009152261989081867,
341
+ "action_probability_max_abs_error": 0.0,
342
+ "argmax_agreements": 3,
343
+ "public_result_equal": false
344
+ },
345
+ {
346
+ "case": "ru",
347
+ "questions": 3,
348
+ "sequence_length": 85,
349
+ "input_sha256": "bac6676f433fdbb4aa2b71dda957299b7e5a9efa2c2b526f000f51d8d4be3a08",
350
+ "tokens_equal": true,
351
+ "logits_max_abs_error": 0.005573272705078125,
352
+ "action_logits_max_abs_error": 2.29296875,
353
+ "probability_max_abs_error": 0.000573146121867707,
354
+ "action_probability_max_abs_error": 0.0,
355
+ "argmax_agreements": 3,
356
+ "public_result_equal": false
357
+ },
358
+ {
359
+ "case": "empty_state",
360
+ "questions": 3,
361
+ "sequence_length": 46,
362
+ "input_sha256": "85f8a747931b65f225679d74c32afc65b23c679549560503921a969453d2c4e1",
363
+ "tokens_equal": true,
364
+ "logits_max_abs_error": 0.0027773380279541016,
365
+ "action_logits_max_abs_error": 2.47705078125,
366
+ "probability_max_abs_error": 0.00039989655194616613,
367
+ "action_probability_max_abs_error": 0.0,
368
+ "argmax_agreements": 3,
369
+ "public_result_equal": false
370
+ },
371
+ {
372
+ "case": "long",
373
+ "questions": 3,
374
+ "sequence_length": 512,
375
+ "input_sha256": "029a742cb223bbb34add55cb46a0d266345a2e448f314ced18033f3710a9c8c7",
376
+ "tokens_equal": true,
377
+ "logits_max_abs_error": 0.0040493011474609375,
378
+ "action_logits_max_abs_error": 2.482421875,
379
+ "probability_max_abs_error": 0.00038620949119971826,
380
+ "action_probability_max_abs_error": 0.0,
381
+ "argmax_agreements": 3,
382
+ "public_result_equal": false
383
+ },
384
+ {
385
+ "case": "conversation",
386
+ "questions": 3,
387
+ "sequence_length": 72,
388
+ "input_sha256": "a429d0393a0bd2af5d960fa45ab286d6741b89969693fa852560a6c87309a4d0",
389
+ "tokens_equal": true,
390
+ "logits_max_abs_error": 0.004204750061035156,
391
+ "action_logits_max_abs_error": 2.8369140625,
392
+ "probability_max_abs_error": 0.001078521392323395,
393
+ "action_probability_max_abs_error": 0.0,
394
+ "argmax_agreements": 3,
395
+ "public_result_equal": false
396
+ },
397
+ {
398
+ "case": "mask_literals",
399
+ "questions": 3,
400
+ "sequence_length": 55,
401
+ "input_sha256": "f89667d91cd59690e55b05e6af9368ea1e27ef42829525a97d4cd3cace2f1989",
402
+ "tokens_equal": true,
403
+ "logits_max_abs_error": 0.020479202270507812,
404
+ "action_logits_max_abs_error": 2.844482421875,
405
+ "probability_max_abs_error": 0.0006111239970967852,
406
+ "action_probability_max_abs_error": 0.0,
407
+ "argmax_agreements": 3,
408
+ "public_result_equal": false
409
+ },
410
+ {
411
+ "case": "many_questions",
412
+ "questions": 20,
413
+ "sequence_length": 93,
414
+ "input_sha256": "47942741e9ed5fabc4c49464b633b43608c61d04c9bc3f23fea5c0ae4da23ae3",
415
+ "tokens_equal": true,
416
+ "logits_max_abs_error": 0.0018677711486816406,
417
+ "action_logits_max_abs_error": 5.8330078125,
418
+ "probability_max_abs_error": 0.00016821344275153027,
419
+ "action_probability_max_abs_error": 0.0,
420
+ "argmax_agreements": 20,
421
+ "public_result_equal": false
422
+ },
423
+ {
424
+ "case": "structured",
425
+ "questions": 3,
426
+ "sequence_length": 78,
427
+ "input_sha256": "8ba1aac3447918529ea99ae8c34146ebe16db359e7787412e6dda90c34f9b8b2",
428
+ "tokens_equal": true,
429
+ "logits_max_abs_error": 0.0077322423458099365,
430
+ "action_logits_max_abs_error": 1.15966796875,
431
+ "probability_max_abs_error": 0.0005516764917031236,
432
+ "action_probability_max_abs_error": 0.0,
433
+ "argmax_agreements": 3,
434
+ "public_result_equal": false
435
+ },
436
+ {
437
+ "case": "twenty_options",
438
+ "questions": 1,
439
+ "sequence_length": 137,
440
+ "input_sha256": "f20fe7490e99ad43d4d1c5a9657f095968b904c9a60e8b3e02fce0544126f838",
441
+ "tokens_equal": true,
442
+ "logits_max_abs_error": 0.005983829498291016,
443
+ "action_logits_max_abs_error": 1.384765625,
444
+ "probability_max_abs_error": 1.541733241969172e-31,
445
+ "action_probability_max_abs_error": 0.0,
446
+ "argmax_agreements": 1,
447
+ "public_result_equal": true
448
+ }
449
+ ],
450
+ "stability": {
451
+ "calls": 100,
452
+ "elapsed_seconds": 3.521159500000067,
453
+ "finite": true,
454
+ "deterministic": true,
455
+ "active_bytes_before": 842602138,
456
+ "active_bytes_after": 842602138,
457
+ "active_growth_bytes": 0,
458
+ "peak_bytes": 1648506739
459
+ }
460
+ }
461
+ ]
462
+ }