Skip to content

Commit a35fbdb

Browse files
mht-sharmadrbhNarsil
authored
Bug Fix: Sliding Window Attention (#3112)
* (fix) sliding window attention * (fix) flashinfer * (typo) collection link * Add window_size_left param ipex rocm * Update window size rocm flash decoding * fix: bump snapshots and improve exceed window test case * feat: add tests for image types and remove alpha from png * Upgrading `from_env` to get token from file when necessary + fix pali_gemma. * fix: add pillow dependency and bump lock+requirements * fix: bump org name in gemma3 test * Fix qwen2. --------- Co-authored-by: drbh <david.richard.holtz@gmail.com> Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
1 parent 8c2c348 commit a35fbdb

33 files changed

+587
-313
lines changed

docs/source/supported_models.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -14,8 +14,8 @@ Text Generation Inference enables serving optimized models. The following sectio
1414
- [Gemma](https://huggingface.co/google/gemma-7b)
1515
- [PaliGemma](https://huggingface.co/google/paligemma-3b-pt-224)
1616
- [Gemma2](https://huggingface.co/collections/google/gemma-2-release-667d6600fd5220e7b967f315)
17-
- [Gemma3](https://huggingface.co/collections/google/gemma-3)
18-
- [Gemma3 Text](https://huggingface.co/collections/google/gemma-3)
17+
- [Gemma3](https://huggingface.co/collections/google/gemma-3-release-67c6c6f89c4f76621268bb6d)
18+
- [Gemma3 Text](https://huggingface.co/collections/google/gemma-3-release-67c6c6f89c4f76621268bb6d)
1919
- [Cohere](https://huggingface.co/CohereForAI/c4ai-command-r-plus)
2020
- [Dbrx](https://huggingface.co/databricks/dbrx-instruct)
2121
- [Mamba](https://huggingface.co/state-spaces/mamba-2.8b-slimpj)
Lines changed: 50 additions & 74 deletions
Original file line numberDiff line numberDiff line change
@@ -1,133 +1,109 @@
11
{
22
"details": {
33
"best_of_sequences": null,
4-
"finish_reason": "length",
5-
"generated_tokens": 20,
4+
"finish_reason": "eos_token",
5+
"generated_tokens": 16,
66
"prefill": [],
77
"seed": null,
88
"tokens": [
99
{
10-
"id": 236764,
11-
"logprob": -0.44726562,
10+
"id": 506,
11+
"logprob": -1.3984375,
1212
"special": false,
13-
"text": ","
13+
"text": " the"
1414
},
1515
{
16-
"id": 236743,
17-
"logprob": -0.011413574,
16+
"id": 1331,
17+
"logprob": -1.6953125,
1818
"special": false,
19-
"text": " "
20-
},
21-
{
22-
"id": 236812,
23-
"logprob": -0.09814453,
24-
"special": false,
25-
"text": "4"
19+
"text": " people"
2620
},
2721
{
2822
"id": 236764,
29-
"logprob": -0.044189453,
23+
"logprob": -0.23535156,
3024
"special": false,
3125
"text": ","
3226
},
3327
{
34-
"id": 236743,
35-
"logprob": -0.15625,
28+
"id": 532,
29+
"logprob": -0.24316406,
3630
"special": false,
37-
"text": " "
31+
"text": " and"
3832
},
3933
{
40-
"id": 236810,
41-
"logprob": -0.010864258,
34+
"id": 506,
35+
"logprob": -0.12109375,
4236
"special": false,
43-
"text": "5"
37+
"text": " the"
4438
},
4539
{
46-
"id": 236764,
47-
"logprob": -0.040039062,
40+
"id": 2780,
41+
"logprob": -1.1640625,
4842
"special": false,
49-
"text": ","
43+
"text": " food"
5044
},
5145
{
52-
"id": 236743,
53-
"logprob": -0.26757812,
46+
"id": 236761,
47+
"logprob": -0.21386719,
5448
"special": false,
55-
"text": " "
49+
"text": "."
5650
},
5751
{
58-
"id": 236825,
59-
"logprob": -0.0047302246,
52+
"id": 108,
53+
"logprob": -0.64453125,
6054
"special": false,
61-
"text": "6"
55+
"text": "\n\n"
6256
},
6357
{
64-
"id": 236764,
65-
"logprob": -0.026123047,
58+
"id": 2094,
59+
"logprob": -0.77734375,
6660
"special": false,
67-
"text": ","
61+
"text": "This"
6862
},
6963
{
70-
"id": 236743,
71-
"logprob": -0.265625,
64+
"id": 563,
65+
"logprob": -0.040283203,
7266
"special": false,
73-
"text": " "
67+
"text": " is"
7468
},
7569
{
76-
"id": 236832,
77-
"logprob": -0.014160156,
70+
"id": 496,
71+
"logprob": -0.03125,
7872
"special": false,
79-
"text": "7"
73+
"text": " a"
8074
},
8175
{
82-
"id": 236764,
83-
"logprob": -0.013977051,
76+
"id": 6290,
77+
"logprob": -0.03515625,
8478
"special": false,
85-
"text": ","
79+
"text": " nice"
8680
},
8781
{
88-
"id": 236743,
89-
"logprob": -0.103515625,
82+
"id": 1977,
83+
"logprob": -0.0020751953,
9084
"special": false,
91-
"text": " "
85+
"text": " place"
9286
},
9387
{
94-
"id": 236828,
95-
"logprob": -0.008178711,
88+
"id": 236761,
89+
"logprob": -0.0079956055,
9690
"special": false,
97-
"text": "8"
91+
"text": "."
9892
},
9993
{
100-
"id": 236764,
101-
"logprob": -0.030151367,
94+
"id": 107,
95+
"logprob": -0.9921875,
10296
"special": false,
103-
"text": ","
97+
"text": "\n"
10498
},
10599
{
106-
"id": 236743,
107-
"logprob": -0.39453125,
108-
"special": false,
109-
"text": " "
110-
},
111-
{
112-
"id": 236819,
113-
"logprob": -0.008728027,
114-
"special": false,
115-
"text": "9"
116-
},
117-
{
118-
"id": 236764,
119-
"logprob": -0.020629883,
120-
"special": false,
121-
"text": ","
122-
},
123-
{
124-
"id": 236743,
125-
"logprob": -0.08154297,
126-
"special": false,
127-
"text": " "
100+
"id": 106,
101+
"logprob": -0.45507812,
102+
"special": true,
103+
"text": "<end_of_turn>"
128104
}
129105
],
130106
"top_tokens": null
131107
},
132-
"generated_text": ", 4, 5, 6, 7, 8, 9, "
108+
"generated_text": " the people, and the food.\n\nThis is a nice place.\n"
133109
}

0 commit comments

Comments
 (0)