File size: 11,174 Bytes
b97f6e6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
import json
import random
import string
import time

import numpy as np
import tritonclient.grpc as client_util
from tokenizers import Tokenizer
from tritonclient.utils import np_to_triton_dtype, InferenceServerException

np.finfo(np.dtype("float32"))
np.finfo(np.dtype("float64"))


class CodeGenProxy:
    def __init__(self, host: str = 'triton', port: int = 8001, verbose: bool = False):
        self.tokenizer = Tokenizer.from_file('/python-docker/cgtok/tokenizer.json')
        self.client = client_util.InferenceServerClient(url=f'{host}:{port}', verbose=verbose)
        self.PAD_CHAR = 50256

        # Max number of tokens the model can handle
        self.MAX_MODEL_LEN = 2048

    class TokensExceedsMaximum(Exception):
        pass

    @staticmethod
    def prepare_tensor(name: str, tensor_input):
        t = client_util.InferInput(
            name, tensor_input.shape, np_to_triton_dtype(tensor_input.dtype))
        t.set_data_from_numpy(tensor_input)
        return t

    @staticmethod
    def trim_with_stopwords(output: str, stopwords: list) -> str:
        for w in sorted(stopwords, key=len, reverse=True):
            if output.endswith(w):
                output = output[:-len(w)]
                break
        return output

    @staticmethod
    def to_word_list_format(word_dict, tokenizer):
        flat_ids = []
        offsets = []
        for word_dict_item in word_dict:
            item_flat_ids = []
            item_offsets = []

            for word in word_dict_item:
                ids = tokenizer.encode(word).ids

                if len(ids) == 0:
                    continue

                item_flat_ids += ids
                item_offsets.append(len(ids))

                # Hack, can we do this better?
                if word == '\n\n':
                    item_flat_ids += [198, 198]
                    item_offsets.append(2)

            flat_ids.append(np.array(item_flat_ids))
            offsets.append(np.cumsum(np.array(item_offsets)))

        pad_to = max(1, max(len(ids) for ids in flat_ids))

        for i, (ids, offs) in enumerate(zip(flat_ids, offsets)):
            flat_ids[i] = np.pad(ids, (0, pad_to - len(ids)), constant_values=0)
            offsets[i] = np.pad(offs, (0, pad_to - len(offs)), constant_values=-1)

        return np.array([flat_ids, offsets], dtype="int32").transpose((1, 0, 2))

    def generate(self, data):
        prompt = data['prompt']
        n = data.get('n', 1)
        model_name = data["model"]
        # ugly hack to set the data type correctly. Huggingface models want int32, but fastertransformer needs uint32
        # i could've done the conversion from uint32 to int32 in the model but that'd be inefficient.
        np_type = np.int32 if model_name.startswith("py-") else np.uint32

        input_start_ids = np.expand_dims(self.tokenizer.encode(prompt).ids, 0)
        input_start_ids = np.repeat(input_start_ids, n, axis=0).astype(np_type)
        prompt_len = input_start_ids.shape[1]
        input_len = prompt_len * np.ones([input_start_ids.shape[0], 1]).astype(np_type)
        max_tokens = data.get('max_tokens', 16)
        prompt_tokens: int = input_len[0][0]
        requested_tokens = max_tokens + prompt_tokens
        if requested_tokens > self.MAX_MODEL_LEN:
            print(1)
            raise self.TokensExceedsMaximum(
                f"This model's maximum context length is {self.MAX_MODEL_LEN}, however you requested "
                f"{requested_tokens} tokens ({prompt_tokens} in your prompt; {max_tokens} for the completion). "
                f"Please reduce your prompt; or completion length."
            )
        output_len = np.ones_like(input_len).astype(np_type) * max_tokens
        num_logprobs = data.get('logprobs', -1)
        if num_logprobs is None:
            num_logprobs = 1
        want_logprobs = num_logprobs > 0

        temperature = data.get('temperature', 0.2)
        if temperature == 0.0:
            temperature = 1.0
            top_k = 1
        else:
            top_k = data.get('top_k', 0)

        top_p = data.get('top_p', 1.0)
        frequency_penalty = data.get('frequency_penalty', 1.0)
        runtime_top_k = top_k * np.ones([input_start_ids.shape[0], 1]).astype(np_type)
        runtime_top_p = top_p * np.ones([input_start_ids.shape[0], 1]).astype(np.float32)
        beam_search_diversity_rate = 0.0 * np.ones([input_start_ids.shape[0], 1]).astype(np.float32)
        random_seed = np.random.randint(0, 2 ** 31 - 1, (input_start_ids.shape[0], 1), dtype=np.int32)
        temperature = temperature * np.ones([input_start_ids.shape[0], 1]).astype(np.float32)
        len_penalty = 1.0 * np.ones([input_start_ids.shape[0], 1]).astype(np.float32)
        repetition_penalty = frequency_penalty * np.ones([input_start_ids.shape[0], 1]).astype(np.float32)
        is_return_log_probs = want_logprobs * np.ones([input_start_ids.shape[0], 1]).astype(np.bool_)
        beam_width = (1 * np.ones([input_start_ids.shape[0], 1])).astype(np_type)
        start_ids = self.PAD_CHAR * np.ones([input_start_ids.shape[0], 1]).astype(np_type)
        end_ids = self.PAD_CHAR * np.ones([input_start_ids.shape[0], 1]).astype(np_type)

        stop_words = data.get('stop', [])
        if stop_words is None:
            stop_words = []
        if stop_words:
            stop_word_list = np.repeat(self.to_word_list_format([stop_words], self.tokenizer), input_start_ids.shape[0],
                                       axis=0)
        else:
            stop_word_list = np.concatenate([np.zeros([input_start_ids.shape[0], 1, 1]).astype(
                np.int32), (-1 * np.ones([input_start_ids.shape[0], 1, 1])).astype(np.int32)], axis=1)

        # Not used
        bad_words_list = np.concatenate([np.zeros([input_start_ids.shape[0], 1, 1]).astype(
            np.int32), (-1 * np.ones([input_start_ids.shape[0], 1, 1])).astype(np.int32)], axis=1)

        inputs = [
            self.prepare_tensor("input_ids", input_start_ids),
            self.prepare_tensor("input_lengths", input_len),
            self.prepare_tensor("request_output_len", output_len),
            self.prepare_tensor("runtime_top_k", runtime_top_k),
            self.prepare_tensor("runtime_top_p", runtime_top_p),
            self.prepare_tensor("beam_search_diversity_rate", beam_search_diversity_rate),
            self.prepare_tensor("random_seed", random_seed),
            self.prepare_tensor("temperature", temperature),
            self.prepare_tensor("len_penalty", len_penalty),
            self.prepare_tensor("repetition_penalty", repetition_penalty),
            self.prepare_tensor("is_return_log_probs", is_return_log_probs),
            self.prepare_tensor("beam_width", beam_width),
            self.prepare_tensor("start_id", start_ids),
            self.prepare_tensor("end_id", end_ids),
            self.prepare_tensor("bad_words_list", bad_words_list),
            self.prepare_tensor("stop_words_list", stop_word_list),
        ]

        result = self.client.infer(model_name, inputs)

        output_data = result.as_numpy("output_ids")
        if output_data is None:
            raise RuntimeError("No output data")

        # All of these squeeze(1)s are to remove the beam width dimension.
        output_data = output_data.squeeze(1)
        if want_logprobs:
            lp_data = result.as_numpy("output_log_probs").squeeze(1)
            # clp_data = result.as_numpy("cum_log_probs").squeeze(1)
        else:
            lp_data = [None] * output_data.shape[0]
        sequence_lengths = result.as_numpy("sequence_length").squeeze(1)
        gen_len = sequence_lengths - input_len.squeeze(1)

        decoded = self.tokenizer.decode_batch([out[prompt_len:prompt_len + g] for g, out in zip(gen_len, output_data)])
        trimmed = [self.trim_with_stopwords(d, stop_words) for d in decoded]

        choices = []
        for i, (text, tokens, lps, g) in enumerate(zip(trimmed, output_data, lp_data, gen_len)):
            reason = "length" if max_tokens == g else "stop"
            if lps is not None:
                tokens_str = [self.tokenizer.decode([t]) for t in tokens[prompt_len:prompt_len + g]]
                offsets = [len(prompt)] + (np.cumsum([len(t) for t in tokens_str]) + len(prompt)).tolist()[:-1]

                # Fake some log probs for top_logprobs
                top_logprobs = []
                for ii, t in enumerate(tokens_str):
                    fakedict = {}
                    top_token_lp = float(lps[ii])
                    fakedict[t] = top_token_lp
                    while len(fakedict) < num_logprobs:
                        random_token = random.randint(0, self.tokenizer.get_vocab_size() - 1)
                        random_token_str = self.tokenizer.decode([random_token])
                        if random_token_str in fakedict:
                            continue
                        random_token_lp = top_token_lp - random.random()
                        fakedict[random_token_str] = random_token_lp
                    top_logprobs.append(fakedict)

                lpdict = {
                    'token_logprobs': lps.tolist(),
                    'top_logprobs': top_logprobs,
                    'tokens': tokens_str,
                    'text_offset': offsets,
                }
            else:
                lpdict = None

            choice = {
                'text': text,
                'index': i,
                'finish_reason': reason,
                'logprobs': lpdict,
            }
            choices.append(choice)

        completion = {
            'id': None,  # fill in
            'model': 'codegen',
            'object': 'text_completion',
            'created': int(time.time()),
            'choices': None,  # fill in
            'usage': {
                'completion_tokens': int(gen_len.sum()),
                'prompt_tokens': int(prompt_len),
                'total_tokens': int(gen_len.sum() + prompt_len),
            }
        }
        return completion, choices

    @staticmethod
    def random_completion_id():
        return 'cmpl-' + ''.join(random.choice(string.ascii_letters + string.digits) for _ in range(29))

    def streamed_response(self, completion, choices):
        for c in choices:
            completion['id'] = self.random_completion_id()
            completion['choices'] = [c]
            yield f'data: {json.dumps(completion)}\n\n'
        yield 'data: [DONE]\n\n'

    def non_streamed_response(self, completion, choices) -> str:
        completion['id'] = self.random_completion_id()
        completion['choices'] = choices
        return json.dumps(completion)

    def __call__(self, data: dict):
        st = time.time()
        try:
            completion, choices = self.generate(data)
        except InferenceServerException as E:
            print(E)
            completion = {}
            choices = []
        ed = time.time()
        print(f"Returned completion in {(ed - st) * 1000} ms")
        if data.get('stream', False):
            return self.streamed_response(completion, choices)
        else:
            return self.non_streamed_response(completion, choices)