Christoph Holthaus
commited on
Commit
·
3c5e66e
1
Parent(s):
878d5c0
dev - magic
Browse files
app.py
CHANGED
|
@@ -64,14 +64,6 @@ MAX_MAX_NEW_TOKENS = 2048
|
|
| 64 |
DEFAULT_MAX_NEW_TOKENS = 1024
|
| 65 |
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
| 66 |
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
if torch.cuda.is_available():
|
| 71 |
-
model_id = "mistralai/Mistral-7B-Instruct-v0.1"
|
| 72 |
-
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float16, device_map="auto")
|
| 73 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 74 |
-
|
| 75 |
# we need to make sure we only run one thread or we probably run out of ram
|
| 76 |
def generate(
|
| 77 |
message: str,
|
|
@@ -87,34 +79,17 @@ def generate(
|
|
| 87 |
conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
|
| 88 |
conversation.append({"role": "user", "content": message})
|
| 89 |
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
llm.generate('test')
|
| 97 |
-
streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
|
| 98 |
-
generate_kwargs = dict(
|
| 99 |
-
{"input_ids": input_ids},
|
| 100 |
-
streamer=streamer,
|
| 101 |
-
max_new_tokens=max_new_tokens,
|
| 102 |
-
do_sample=True,
|
| 103 |
-
top_p=top_p,
|
| 104 |
-
top_k=top_k,
|
| 105 |
-
temperature=temperature,
|
| 106 |
-
num_beams=1,
|
| 107 |
-
repetition_penalty=repetition_penalty,
|
| 108 |
-
)
|
| 109 |
-
t = Thread(target=model.generate, kwargs=generate_kwargs)
|
| 110 |
-
t.start()
|
| 111 |
|
| 112 |
outputs = []
|
| 113 |
for text in streamer:
|
| 114 |
outputs.append(text)
|
| 115 |
yield "".join(outputs)
|
| 116 |
|
| 117 |
-
|
| 118 |
chat_interface = gr.ChatInterface(
|
| 119 |
fn=generate,
|
| 120 |
additional_inputs=[
|
|
|
|
| 64 |
DEFAULT_MAX_NEW_TOKENS = 1024
|
| 65 |
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
|
| 66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
# we need to make sure we only run one thread or we probably run out of ram
|
| 68 |
def generate(
|
| 69 |
message: str,
|
|
|
|
| 79 |
conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
|
| 80 |
conversation.append({"role": "user", "content": message})
|
| 81 |
|
| 82 |
+
# Use LLaMa to create chat completion
|
| 83 |
+
llm.create_chat_completion(conversation, stream=True)
|
| 84 |
+
|
| 85 |
+
# Initialize a TextIteratorStreamer
|
| 86 |
+
streamer = TextIteratorStreamer(llm, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
|
| 88 |
outputs = []
|
| 89 |
for text in streamer:
|
| 90 |
outputs.append(text)
|
| 91 |
yield "".join(outputs)
|
| 92 |
|
|
|
|
| 93 |
chat_interface = gr.ChatInterface(
|
| 94 |
fn=generate,
|
| 95 |
additional_inputs=[
|