Spaces:
Running
on
Zero
Running
on
Zero
Commit
·
eb1a863
1
Parent(s):
3ff8c65
reactivate gemma models, add some ice cream, support peft
Browse files- requirements.txt +2 -1
- utils/models.py +156 -66
- utils/prompts.py +1 -1
requirements.txt
CHANGED
@@ -7,4 +7,5 @@ openai>=1.60.2
|
|
7 |
torch>=2.5.1
|
8 |
tqdm==4.67.1
|
9 |
vllm>=0.8.5
|
10 |
-
spaces
|
|
|
|
7 |
torch>=2.5.1
|
8 |
tqdm==4.67.1
|
9 |
vllm>=0.8.5
|
10 |
+
spaces
|
11 |
+
peft>=0.15.1
|
utils/models.py
CHANGED
@@ -1,32 +1,41 @@
|
|
1 |
import os
|
2 |
-
|
|
|
3 |
import spaces
|
|
|
|
|
4 |
|
5 |
import torch
|
6 |
-
from transformers import
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
from .prompts import format_rag_prompt
|
8 |
from .shared import generation_interrupt
|
9 |
|
10 |
models = {
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
"Cogito-v1-preview-llama-3b": "deepcogito/cogito-v1-preview-llama-3b",
|
20 |
-
|
21 |
# #"Bitnet-b1.58-2B4T": "microsoft/bitnet-b1.58-2B-4T",
|
22 |
# #"MiniCPM3-RAG-LoRA": "openbmb/MiniCPM3-RAG-LoRA",
|
23 |
"Qwen3-0.6b": "qwen/qwen3-0.6b",
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
}
|
31 |
|
32 |
tokenizer_cache = {}
|
@@ -34,14 +43,16 @@ tokenizer_cache = {}
|
|
34 |
# List of model names for easy access
|
35 |
model_names = list(models.keys())
|
36 |
|
|
|
37 |
# Custom stopping criteria that checks the interrupt flag
|
38 |
class InterruptCriteria(StoppingCriteria):
|
39 |
def __init__(self, interrupt_event):
|
40 |
self.interrupt_event = interrupt_event
|
41 |
-
|
42 |
def __call__(self, input_ids, scores, **kwargs):
|
43 |
return self.interrupt_event.is_set()
|
44 |
|
|
|
45 |
@spaces.GPU
|
46 |
def generate_summaries(example, model_a_name, model_b_name):
|
47 |
"""
|
@@ -49,48 +60,49 @@ def generate_summaries(example, model_a_name, model_b_name):
|
|
49 |
"""
|
50 |
if generation_interrupt.is_set():
|
51 |
return "", ""
|
52 |
-
|
53 |
context_text = ""
|
54 |
context_parts = []
|
55 |
-
|
56 |
if "full_contexts" in example and example["full_contexts"]:
|
57 |
for i, ctx in enumerate(example["full_contexts"]):
|
58 |
content = ""
|
59 |
-
|
60 |
# Extract content from either dict or string
|
61 |
if isinstance(ctx, dict) and "content" in ctx:
|
62 |
content = ctx["content"]
|
63 |
elif isinstance(ctx, str):
|
64 |
content = ctx
|
65 |
-
|
66 |
# Add document number if not already present
|
67 |
if not content.strip().startswith("Document"):
|
68 |
-
content = f"Document {i+1}:\n{content}"
|
69 |
-
|
70 |
context_parts.append(content)
|
71 |
-
|
72 |
context_text = "\n\n".join(context_parts)
|
73 |
else:
|
74 |
# Provide a graceful fallback instead of raising an error
|
75 |
print("Warning: No full context found in the example, using empty context")
|
76 |
context_text = ""
|
77 |
-
|
78 |
question = example.get("question", "")
|
79 |
-
|
80 |
if generation_interrupt.is_set():
|
81 |
return "", ""
|
82 |
-
|
83 |
# Run model A
|
84 |
summary_a = run_inference(models[model_a_name], context_text, question)
|
85 |
-
|
86 |
if generation_interrupt.is_set():
|
87 |
return summary_a, ""
|
88 |
-
|
89 |
# Run model B
|
90 |
summary_b = run_inference(models[model_b_name], context_text, question)
|
91 |
-
|
92 |
return summary_a, summary_b
|
93 |
|
|
|
94 |
@spaces.GPU
|
95 |
def run_inference(model_name, context, question):
|
96 |
"""
|
@@ -105,29 +117,40 @@ def run_inference(model_name, context, question):
|
|
105 |
result = ""
|
106 |
tokenizer_kwargs = {
|
107 |
"add_generation_prompt": True,
|
108 |
-
}
|
109 |
generation_kwargs = {
|
110 |
"max_new_tokens": 512,
|
111 |
}
|
112 |
-
if "qwen3" in model_name.lower():
|
113 |
-
print(
|
|
|
|
|
114 |
tokenizer_kwargs["enable_thinking"] = False
|
115 |
|
116 |
try:
|
|
|
117 |
if model_name in tokenizer_cache:
|
118 |
tokenizer = tokenizer_cache[model_name]
|
119 |
else:
|
120 |
-
tokenizer
|
121 |
-
|
122 |
-
padding_side="left",
|
123 |
-
token=True,
|
124 |
-
kwargs=tokenizer_kwargs
|
125 |
-
)
|
126 |
-
tokenizer_cache[model_name] = tokenizer
|
127 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
128 |
accepts_sys = (
|
129 |
"System role not supported" not in tokenizer.chat_template
|
130 |
-
if tokenizer.chat_template
|
|
|
131 |
)
|
132 |
|
133 |
if tokenizer.pad_token is None:
|
@@ -136,40 +159,107 @@ def run_inference(model_name, context, question):
|
|
136 |
# Check interrupt before loading the model
|
137 |
if generation_interrupt.is_set():
|
138 |
return ""
|
139 |
-
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
152 |
|
153 |
text_input = format_rag_prompt(question, context, accepts_sys)
|
154 |
-
if "Gemma-3".lower()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
155 |
formatted = pipe.tokenizer.apply_chat_template(
|
156 |
text_input,
|
157 |
-
tokenize=
|
158 |
**tokenizer_kwargs,
|
159 |
)
|
160 |
-
|
161 |
input_length = len(formatted)
|
162 |
-
|
163 |
|
164 |
-
outputs = pipe(
|
165 |
-
|
166 |
-
|
167 |
-
|
168 |
-
|
169 |
-
|
|
|
170 |
|
171 |
except Exception as e:
|
172 |
print(f"Error in inference for {model_name}: {e}")
|
|
|
173 |
result = f"Error generating response: {str(e)[:200]}..."
|
174 |
|
175 |
finally:
|
@@ -177,4 +267,4 @@ def run_inference(model_name, context, question):
|
|
177 |
if torch.cuda.is_available():
|
178 |
torch.cuda.empty_cache()
|
179 |
|
180 |
-
return result
|
|
|
1 |
import os
|
2 |
+
|
3 |
+
os.environ["MKL_THREADING_LAYER"] = "GNU"
|
4 |
import spaces
|
5 |
+
from peft import PeftModel
|
6 |
+
import traceback
|
7 |
|
8 |
import torch
|
9 |
+
from transformers import (
|
10 |
+
pipeline,
|
11 |
+
AutoTokenizer,
|
12 |
+
AutoModelForCausalLM,
|
13 |
+
StoppingCriteria,
|
14 |
+
StoppingCriteriaList,
|
15 |
+
)
|
16 |
from .prompts import format_rag_prompt
|
17 |
from .shared import generation_interrupt
|
18 |
|
19 |
models = {
|
20 |
+
"Qwen2.5-1.5b-Instruct": "qwen/qwen2.5-1.5b-instruct",
|
21 |
+
"Qwen2.5-3b-Instruct": "qwen/qwen2.5-3b-instruct",
|
22 |
+
"Llama-3.2-1b-Instruct": "meta-llama/llama-3.2-1b-instruct",
|
23 |
+
"Llama-3.2-3b-Instruct": "meta-llama/llama-3.2-3b-instruct",
|
24 |
+
"Gemma-3-1b-it": "google/gemma-3-1b-it",
|
25 |
+
"Gemma-3-4b-it": "google/gemma-3-4b-it",
|
26 |
+
"Gemma-2-2b-it": "google/gemma-2-2b-it",
|
27 |
+
"Phi-4-mini-instruct": "microsoft/phi-4-mini-instruct",
|
28 |
"Cogito-v1-preview-llama-3b": "deepcogito/cogito-v1-preview-llama-3b",
|
29 |
+
"IBM Granite-3.3-2b-instruct": "ibm-granite/granite-3.3-2b-instruct",
|
30 |
# #"Bitnet-b1.58-2B4T": "microsoft/bitnet-b1.58-2B-4T",
|
31 |
# #"MiniCPM3-RAG-LoRA": "openbmb/MiniCPM3-RAG-LoRA",
|
32 |
"Qwen3-0.6b": "qwen/qwen3-0.6b",
|
33 |
+
"Qwen3-1.7b": "qwen/qwen3-1.7b",
|
34 |
+
"Qwen3-4b": "qwen/qwen3-4b",
|
35 |
+
"SmolLM2-1.7b-Instruct": "HuggingFaceTB/SmolLM2-1.7B-Instruct",
|
36 |
+
"EXAONE-3.5-2.4B-instruct": "LGAI-EXAONE/EXAONE-3.5-2.4B-Instruct",
|
37 |
+
"OLMo-2-1B-Instruct": "allenai/OLMo-2-0425-1B-Instruct",
|
38 |
+
"icecream-3b": "aizip-dev/icecream-3b",
|
39 |
}
|
40 |
|
41 |
tokenizer_cache = {}
|
|
|
43 |
# List of model names for easy access
|
44 |
model_names = list(models.keys())
|
45 |
|
46 |
+
|
47 |
# Custom stopping criteria that checks the interrupt flag
|
48 |
class InterruptCriteria(StoppingCriteria):
|
49 |
def __init__(self, interrupt_event):
|
50 |
self.interrupt_event = interrupt_event
|
51 |
+
|
52 |
def __call__(self, input_ids, scores, **kwargs):
|
53 |
return self.interrupt_event.is_set()
|
54 |
|
55 |
+
|
56 |
@spaces.GPU
|
57 |
def generate_summaries(example, model_a_name, model_b_name):
|
58 |
"""
|
|
|
60 |
"""
|
61 |
if generation_interrupt.is_set():
|
62 |
return "", ""
|
63 |
+
|
64 |
context_text = ""
|
65 |
context_parts = []
|
66 |
+
|
67 |
if "full_contexts" in example and example["full_contexts"]:
|
68 |
for i, ctx in enumerate(example["full_contexts"]):
|
69 |
content = ""
|
70 |
+
|
71 |
# Extract content from either dict or string
|
72 |
if isinstance(ctx, dict) and "content" in ctx:
|
73 |
content = ctx["content"]
|
74 |
elif isinstance(ctx, str):
|
75 |
content = ctx
|
76 |
+
|
77 |
# Add document number if not already present
|
78 |
if not content.strip().startswith("Document"):
|
79 |
+
content = f"Document {i + 1}:\n{content}"
|
80 |
+
|
81 |
context_parts.append(content)
|
82 |
+
|
83 |
context_text = "\n\n".join(context_parts)
|
84 |
else:
|
85 |
# Provide a graceful fallback instead of raising an error
|
86 |
print("Warning: No full context found in the example, using empty context")
|
87 |
context_text = ""
|
88 |
+
|
89 |
question = example.get("question", "")
|
90 |
+
|
91 |
if generation_interrupt.is_set():
|
92 |
return "", ""
|
93 |
+
|
94 |
# Run model A
|
95 |
summary_a = run_inference(models[model_a_name], context_text, question)
|
96 |
+
|
97 |
if generation_interrupt.is_set():
|
98 |
return summary_a, ""
|
99 |
+
|
100 |
# Run model B
|
101 |
summary_b = run_inference(models[model_b_name], context_text, question)
|
102 |
+
|
103 |
return summary_a, summary_b
|
104 |
|
105 |
+
|
106 |
@spaces.GPU
|
107 |
def run_inference(model_name, context, question):
|
108 |
"""
|
|
|
117 |
result = ""
|
118 |
tokenizer_kwargs = {
|
119 |
"add_generation_prompt": True,
|
120 |
+
} # make sure qwen3 doesn't use thinking
|
121 |
generation_kwargs = {
|
122 |
"max_new_tokens": 512,
|
123 |
}
|
124 |
+
if "qwen3" in model_name.lower():
|
125 |
+
print(
|
126 |
+
f"Recognized {model_name} as a Qwen3 model. Setting enable_thinking=False."
|
127 |
+
)
|
128 |
tokenizer_kwargs["enable_thinking"] = False
|
129 |
|
130 |
try:
|
131 |
+
print("REACHED HERE BEFORE tokenizer")
|
132 |
if model_name in tokenizer_cache:
|
133 |
tokenizer = tokenizer_cache[model_name]
|
134 |
else:
|
135 |
+
# Common arguments for tokenizer loading
|
136 |
+
tokenizer_load_args = {"padding_side": "left", "token": True}
|
|
|
|
|
|
|
|
|
|
|
137 |
|
138 |
+
# Determine the Hugging Face model name for the tokenizer
|
139 |
+
actual_model_name_for_tokenizer = model_name
|
140 |
+
if "icecream" in model_name.lower():
|
141 |
+
actual_model_name_for_tokenizer = "meta-llama/llama-3.2-3b-instruct"
|
142 |
+
|
143 |
+
# Note: tokenizer_kwargs (defined earlier, with add_generation_prompt etc.)
|
144 |
+
# is intended for tokenizer.apply_chat_template, not for AutoTokenizer.from_pretrained generally.
|
145 |
+
# If a specific tokenizer (e.g., Qwen) needs special __init__ args that happen to be in tokenizer_kwargs,
|
146 |
+
# that would require more specific handling here. For now, we assume general constructor args.
|
147 |
+
tokenizer = AutoTokenizer.from_pretrained(actual_model_name_for_tokenizer, **tokenizer_load_args)
|
148 |
+
tokenizer_cache[model_name] = tokenizer
|
149 |
+
|
150 |
accepts_sys = (
|
151 |
"System role not supported" not in tokenizer.chat_template
|
152 |
+
if tokenizer.chat_template
|
153 |
+
else False # Handle missing chat_template
|
154 |
)
|
155 |
|
156 |
if tokenizer.pad_token is None:
|
|
|
159 |
# Check interrupt before loading the model
|
160 |
if generation_interrupt.is_set():
|
161 |
return ""
|
162 |
+
|
163 |
+
print("REACHED HERE BEFORE pipe")
|
164 |
+
print(f"Loading model {model_name}...")
|
165 |
+
if "icecream" not in model_name.lower():
|
166 |
+
pipe = pipeline(
|
167 |
+
"text-generation",
|
168 |
+
model=model_name,
|
169 |
+
tokenizer=tokenizer,
|
170 |
+
device_map="cuda",
|
171 |
+
trust_remote_code=True,
|
172 |
+
torch_dtype=torch.bfloat16,
|
173 |
+
model_kwargs={
|
174 |
+
"attn_implementation": "eager",
|
175 |
+
},
|
176 |
+
)
|
177 |
+
else:
|
178 |
+
base_model = AutoModelForCausalLM.from_pretrained(
|
179 |
+
"meta-llama/llama-3.2-3b-instruct",
|
180 |
+
device_map="cuda",
|
181 |
+
torch_dtype=torch.bfloat16,
|
182 |
+
trust_remote_code=True,
|
183 |
+
)
|
184 |
+
model = PeftModel.from_pretrained(
|
185 |
+
base_model,
|
186 |
+
"aizip-dev/icecream-3b",
|
187 |
+
device_map="cuda",
|
188 |
+
torch_dtype=torch.bfloat16,
|
189 |
+
)
|
190 |
|
191 |
text_input = format_rag_prompt(question, context, accepts_sys)
|
192 |
+
if "Gemma-3".lower() in model_name.lower():
|
193 |
+
print("REACHED HERE BEFORE GEN")
|
194 |
+
result = pipe(
|
195 |
+
text_input,
|
196 |
+
max_new_tokens=512,
|
197 |
+
generation_kwargs={"skip_special_tokens": True},
|
198 |
+
)[0]["generated_text"]
|
199 |
+
|
200 |
+
result = result[-1]["content"]
|
201 |
+
elif "icecream" in model_name.lower():
|
202 |
+
|
203 |
+
print("ICECREAM")
|
204 |
+
# text_input is the list of messages from format_rag_prompt
|
205 |
+
# tokenizer_kwargs (e.g., {"add_generation_prompt": True}) are correctly passed to apply_chat_template
|
206 |
+
model_inputs = tokenizer.apply_chat_template(
|
207 |
+
text_input,
|
208 |
+
tokenize=True,
|
209 |
+
return_tensors="pt",
|
210 |
+
return_dict=True,
|
211 |
+
**tokenizer_kwargs,
|
212 |
+
)
|
213 |
+
|
214 |
+
# Move all tensors within the BatchEncoding (model_inputs) to the model's device
|
215 |
+
model_inputs = model_inputs.to(model.device)
|
216 |
+
|
217 |
+
input_ids = model_inputs.input_ids
|
218 |
+
attention_mask = model_inputs.attention_mask # Expecting this from a correctly configured tokenizer
|
219 |
+
|
220 |
+
prompt_tokens_length = input_ids.shape[1] # Get length of tokenized prompt
|
221 |
+
|
222 |
+
with torch.inference_mode():
|
223 |
+
# Check interrupt before generation
|
224 |
+
if generation_interrupt.is_set():
|
225 |
+
return ""
|
226 |
+
|
227 |
+
# Explicitly pass input_ids, attention_mask, and pad_token_id
|
228 |
+
# tokenizer.pad_token is set to tokenizer.eos_token if None, earlier in the code.
|
229 |
+
output_sequences = model.generate(
|
230 |
+
input_ids=input_ids,
|
231 |
+
attention_mask=attention_mask,
|
232 |
+
max_new_tokens=512,
|
233 |
+
eos_token_id=tokenizer.eos_token_id, # Good practice for stopping generation
|
234 |
+
pad_token_id=tokenizer.pad_token_id # Addresses the warning
|
235 |
+
)
|
236 |
+
|
237 |
+
# output_sequences[0] contains the full sequence (prompt + generation)
|
238 |
+
# Decode only the newly generated tokens
|
239 |
+
generated_token_ids = output_sequences[0][prompt_tokens_length:]
|
240 |
+
result = tokenizer.decode(generated_token_ids, skip_special_tokens=True)
|
241 |
+
|
242 |
+
else: # For other models
|
243 |
formatted = pipe.tokenizer.apply_chat_template(
|
244 |
text_input,
|
245 |
+
tokenize=True,
|
246 |
**tokenizer_kwargs,
|
247 |
)
|
248 |
+
|
249 |
input_length = len(formatted)
|
250 |
+
# Check interrupt before generation
|
251 |
|
252 |
+
outputs = pipe(
|
253 |
+
formatted,
|
254 |
+
max_new_tokens=512,
|
255 |
+
generation_kwargs={"skip_special_tokens": True},
|
256 |
+
)
|
257 |
+
# print(outputs[0]['generated_text'])
|
258 |
+
result = outputs[0]["generated_text"][input_length:]
|
259 |
|
260 |
except Exception as e:
|
261 |
print(f"Error in inference for {model_name}: {e}")
|
262 |
+
print(traceback.format_exc())
|
263 |
result = f"Error generating response: {str(e)[:200]}..."
|
264 |
|
265 |
finally:
|
|
|
267 |
if torch.cuda.is_available():
|
268 |
torch.cuda.empty_cache()
|
269 |
|
270 |
+
return result
|
utils/prompts.py
CHANGED
@@ -26,7 +26,7 @@ Given the following query and context, please provide your response:
|
|
26 |
|
27 |
{context}
|
28 |
|
29 |
-
WITHOUT mentioning your judgement either your grounded answer, OR refusal and clarifications:
|
30 |
"""
|
31 |
|
32 |
messages = (
|
|
|
26 |
|
27 |
{context}
|
28 |
|
29 |
+
WITHOUT mentioning your judgement on answerability, either your grounded answer, OR refusal and clarifications:
|
30 |
"""
|
31 |
|
32 |
messages = (
|