Spaces:
Sleeping
Sleeping
jed-tiotuico
commited on
Commit
•
d47ab3c
1
Parent(s):
673d242
changed to ft model
Browse files
app.py
CHANGED
@@ -6,16 +6,16 @@ import threading
|
|
6 |
import streamlit as st
|
7 |
import random
|
8 |
from typing import Iterable
|
9 |
-
|
10 |
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer, PreTrainedTokenizerFast
|
11 |
from datetime import datetime
|
12 |
from threading import Thread
|
13 |
|
14 |
-
|
15 |
-
|
|
|
|
|
16 |
|
17 |
-
fine_tuned_model_name = "MBZUAI/LaMini-GPT-124M"
|
18 |
-
sota_model_name = "MBZUAI/LaMini-GPT-124M"
|
19 |
alpaca_input_text_format = "### Instruction:\n{}\n\n### Response:\n"
|
20 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
21 |
# if device is cpu try mps?
|
@@ -24,16 +24,15 @@ if device == "cpu":
|
|
24 |
device = torch.device("mps" if torch.backends.mps.is_available() else "cpu")
|
25 |
|
26 |
def get_model_tokenizer(sota_model_name):
|
27 |
-
tokenizer =
|
28 |
-
|
29 |
-
|
30 |
-
|
|
|
|
|
|
|
31 |
)
|
32 |
-
|
33 |
-
sota_model_name,
|
34 |
-
cache_dir="/data/.hf_cache",
|
35 |
-
trust_remote_code=True
|
36 |
-
).to(device)
|
37 |
|
38 |
return model, tokenizer
|
39 |
|
@@ -61,16 +60,14 @@ def write_stream_user_chat_message(user_chat, model, token, prompt):
|
|
61 |
return new_customer_msg
|
62 |
|
63 |
def get_mistral_model_tokenizer(sota_model_name):
|
64 |
-
tokenizer =
|
65 |
-
|
66 |
-
|
67 |
-
|
|
|
|
|
68 |
)
|
69 |
-
|
70 |
-
sota_model_name,
|
71 |
-
cache_dir="/data/.hf_cache",
|
72 |
-
trust_remote_code=True
|
73 |
-
).to(device)
|
74 |
|
75 |
return model, tokenizer
|
76 |
|
|
|
6 |
import streamlit as st
|
7 |
import random
|
8 |
from typing import Iterable
|
9 |
+
from unsloth import FastLanguageModel
|
10 |
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer, PreTrainedTokenizerFast
|
11 |
from datetime import datetime
|
12 |
from threading import Thread
|
13 |
|
14 |
+
fine_tuned_model_name = "jed-tiotuico/twitter-llama"
|
15 |
+
sota_model_name = "unsloth/mistral-7b-instruct-v0.2-bnb-4bit"
|
16 |
+
# fine_tuned_model_name = "MBZUAI/LaMini-GPT-124M"
|
17 |
+
# sota_model_name = "MBZUAI/LaMini-GPT-124M"
|
18 |
|
|
|
|
|
19 |
alpaca_input_text_format = "### Instruction:\n{}\n\n### Response:\n"
|
20 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
21 |
# if device is cpu try mps?
|
|
|
24 |
device = torch.device("mps" if torch.backends.mps.is_available() else "cpu")
|
25 |
|
26 |
def get_model_tokenizer(sota_model_name):
|
27 |
+
model, tokenizer = FastLanguageModel.from_pretrained(
|
28 |
+
model_name = "jed-tiotuico/twitter-llama",
|
29 |
+
max_seq_length = 200,
|
30 |
+
dtype = None,
|
31 |
+
load_in_4bit = True,
|
32 |
+
cache_dir = "/data/.cache/hf-models",
|
33 |
+
token=st.secrets["HF_TOKEN"]
|
34 |
)
|
35 |
+
FastLanguageModel.for_inference(model)
|
|
|
|
|
|
|
|
|
36 |
|
37 |
return model, tokenizer
|
38 |
|
|
|
60 |
return new_customer_msg
|
61 |
|
62 |
def get_mistral_model_tokenizer(sota_model_name):
|
63 |
+
model, tokenizer = FastLanguageModel.from_pretrained(
|
64 |
+
model_name = "unsloth/mistral-7b-instruct-v0.2-bnb-4bit",
|
65 |
+
max_seq_length = max_seq_length,
|
66 |
+
dtype = dtype,
|
67 |
+
load_in_4bit = load_in_4bit,
|
68 |
+
cache_dir = "/data/.cache/hf-models",
|
69 |
)
|
70 |
+
FastLanguageModel.for_inference(model)
|
|
|
|
|
|
|
|
|
71 |
|
72 |
return model, tokenizer
|
73 |
|