Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -14,7 +14,8 @@ torch.set_num_threads(2)
|
|
14 |
|
15 |
def min_p_sampling(logits, pbase=0.1):
|
16 |
"""
|
17 |
-
Perform min-p sampling on the logits.
|
|
|
18 |
|
19 |
Args:
|
20 |
logits (torch.Tensor): 1D tensor of logits for the next token.
|
@@ -47,6 +48,96 @@ def min_p_sampling(logits, pbase=0.1):
|
|
47 |
return sampled_index.item()
|
48 |
|
49 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
def generate_completion(prompt, strategy, params):
|
51 |
"""
|
52 |
Generate a complete answer using model.generate with specified parameters.
|
@@ -59,12 +150,12 @@ def generate_completion(prompt, strategy, params):
|
|
59 |
|
60 |
# Generate the output.
|
61 |
output_ids = model.generate(
|
62 |
-
input_ids, attention_mask=attention_mask, max_length=
|
63 |
)
|
64 |
return tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
65 |
|
66 |
|
67 |
-
def generate_min_p_completion(prompt, pbase=0.1, max_length=
|
68 |
input_ids = tokenizer.encode(prompt, return_tensors="pt")
|
69 |
past = None
|
70 |
with torch.no_grad():
|
@@ -94,7 +185,7 @@ def generate_all(prompt):
|
|
94 |
"Greedy": {"type": "default", "params": {"do_sample": False}},
|
95 |
"Top-k Sampling": {
|
96 |
"type": "default",
|
97 |
-
"params": {"do_sample": True, "top_k":
|
98 |
},
|
99 |
"Top-p Sampling": {
|
100 |
"type": "default",
|
@@ -113,6 +204,14 @@ def generate_all(prompt):
|
|
113 |
"params": {"do_sample": True, "epsilon_cutoff": 0.2},
|
114 |
},
|
115 |
"Min-p Sampling": {"type": "min_p", "pbase": 0.1},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
116 |
}
|
117 |
|
118 |
# Define the order for display.
|
@@ -124,6 +223,8 @@ def generate_all(prompt):
|
|
124 |
"Min-p Sampling",
|
125 |
"Eta Sampling",
|
126 |
"Epsilon Sampling",
|
|
|
|
|
127 |
]
|
128 |
results = {method: None for method in methods}
|
129 |
|
@@ -142,6 +243,11 @@ def generate_all(prompt):
|
|
142 |
future = executor.submit(
|
143 |
generate_min_p_completion, prompt, info["pbase"]
|
144 |
)
|
|
|
|
|
|
|
|
|
|
|
145 |
future_to_method[future] = method
|
146 |
|
147 |
# As each future completes, update its result and yield the current state.
|
@@ -169,9 +275,15 @@ interface = gr.Interface(
|
|
169 |
gr.Textbox(label="Top-k Sampling"),
|
170 |
gr.Textbox(label="Top-p Sampling"),
|
171 |
gr.Textbox(label="Beam Search"),
|
172 |
-
gr.Textbox(label="Min-p Sampling"),
|
173 |
gr.Textbox(label="Eta Sampling"),
|
174 |
gr.Textbox(label="Epsilon Sampling"),
|
|
|
|
|
|
|
|
|
|
|
|
|
175 |
],
|
176 |
title="Decoding Methods Comparison",
|
177 |
description="Each decoding method's final answer is printed as soon as it is done. Model used: GPT-2.",
|
|
|
14 |
|
15 |
def min_p_sampling(logits, pbase=0.1):
|
16 |
"""
|
17 |
+
Perform min-p sampling on the logits. As described in
|
18 |
+
https://arxiv.org/abs/2407.01082
|
19 |
|
20 |
Args:
|
21 |
logits (torch.Tensor): 1D tensor of logits for the next token.
|
|
|
48 |
return sampled_index.item()
|
49 |
|
50 |
|
51 |
+
def generate_laconic_completion(prompt: str, n: int = 5, max_length: int = 100):
|
52 |
+
# generate n completions greedily and return the shortest one
|
53 |
+
with torch.no_grad():
|
54 |
+
# Encode the prompt and get the attention mask.
|
55 |
+
encoded = tokenizer(prompt, return_tensors="pt")
|
56 |
+
input_ids = encoded["input_ids"]
|
57 |
+
attention_mask = encoded["attention_mask"]
|
58 |
+
|
59 |
+
# Generate the output.
|
60 |
+
outputs = model.generate(
|
61 |
+
input_ids,
|
62 |
+
attention_mask=attention_mask,
|
63 |
+
max_length=max_length,
|
64 |
+
num_return_sequences=n,
|
65 |
+
do_sample=True,
|
66 |
+
)
|
67 |
+
completions = [
|
68 |
+
tokenizer.decode(output, skip_special_tokens=True) for output in outputs
|
69 |
+
]
|
70 |
+
return min(completions, key=len)
|
71 |
+
|
72 |
+
|
73 |
+
def generate_with_confidence(input_ids, max_length):
|
74 |
+
"""
|
75 |
+
Generate a sequence using greedy decoding while returning the scores.
|
76 |
+
"""
|
77 |
+
outputs = model.generate(
|
78 |
+
input_ids,
|
79 |
+
max_length=max_length,
|
80 |
+
do_sample=False,
|
81 |
+
output_scores=True,
|
82 |
+
return_dict_in_generate=True,
|
83 |
+
)
|
84 |
+
return outputs
|
85 |
+
|
86 |
+
|
87 |
+
def compute_answer_confidence(outputs):
|
88 |
+
"""
|
89 |
+
Compute the answer confidence over the generated tokens.
|
90 |
+
For each generated token, compute the difference between the top-1 and top-2 logits.
|
91 |
+
Returns the average difference.
|
92 |
+
"""
|
93 |
+
diffs = []
|
94 |
+
for score in outputs.scores:
|
95 |
+
# Get top-2 logit values
|
96 |
+
top2 = torch.topk(score[0], 2)
|
97 |
+
diff = top2.values[0] - top2.values[1]
|
98 |
+
diffs.append(diff.item())
|
99 |
+
|
100 |
+
return sum(diffs) / len(diffs) if diffs else 0.0
|
101 |
+
|
102 |
+
|
103 |
+
def cot_decoding(prompt, k=5, max_length=100):
|
104 |
+
"""
|
105 |
+
Perform Chain-of-Thought (CoT) decoding by exploring top-k alternative paths.
|
106 |
+
"""
|
107 |
+
input_ids = tokenizer.encode(prompt, return_tensors="pt")
|
108 |
+
|
109 |
+
# Get logits for the next token
|
110 |
+
with torch.no_grad():
|
111 |
+
outputs = model(input_ids)
|
112 |
+
logits = outputs.logits[0, -1, :]
|
113 |
+
|
114 |
+
# Get top-k candidate tokens
|
115 |
+
topk = torch.topk(logits, k)
|
116 |
+
candidate_tokens = topk.indices
|
117 |
+
|
118 |
+
paths = []
|
119 |
+
for token in candidate_tokens:
|
120 |
+
# Append the candidate token to the prompt
|
121 |
+
new_input_ids = torch.cat([input_ids, token.view(1, 1)], dim=1)
|
122 |
+
|
123 |
+
# Generate a full sequence with output scores
|
124 |
+
gen_outputs = generate_with_confidence(
|
125 |
+
new_input_ids, max_length=new_input_ids.shape[1] + max_length
|
126 |
+
)
|
127 |
+
|
128 |
+
# Decode the generated sequence
|
129 |
+
generated_text = tokenizer.decode(
|
130 |
+
gen_outputs.sequences[0], skip_special_tokens=True
|
131 |
+
)
|
132 |
+
|
133 |
+
# Compute answer confidence
|
134 |
+
confidence = compute_answer_confidence(gen_outputs)
|
135 |
+
|
136 |
+
paths.append({"text": generated_text, "confidence": confidence})
|
137 |
+
|
138 |
+
return max(paths, key=lambda x: x["confidence"])["text"]
|
139 |
+
|
140 |
+
|
141 |
def generate_completion(prompt, strategy, params):
|
142 |
"""
|
143 |
Generate a complete answer using model.generate with specified parameters.
|
|
|
150 |
|
151 |
# Generate the output.
|
152 |
output_ids = model.generate(
|
153 |
+
input_ids, attention_mask=attention_mask, max_length=100, **params
|
154 |
)
|
155 |
return tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
156 |
|
157 |
|
158 |
+
def generate_min_p_completion(prompt, pbase=0.1, max_length=100):
|
159 |
input_ids = tokenizer.encode(prompt, return_tensors="pt")
|
160 |
past = None
|
161 |
with torch.no_grad():
|
|
|
185 |
"Greedy": {"type": "default", "params": {"do_sample": False}},
|
186 |
"Top-k Sampling": {
|
187 |
"type": "default",
|
188 |
+
"params": {"do_sample": True, "top_k": 100},
|
189 |
},
|
190 |
"Top-p Sampling": {
|
191 |
"type": "default",
|
|
|
204 |
"params": {"do_sample": True, "epsilon_cutoff": 0.2},
|
205 |
},
|
206 |
"Min-p Sampling": {"type": "min_p", "pbase": 0.1},
|
207 |
+
"laconic": {
|
208 |
+
"type": "default",
|
209 |
+
"params": {"do_sample": True, "num_return_sequences": 5},
|
210 |
+
},
|
211 |
+
"COT Decoding": {
|
212 |
+
"type": "cot_decoding",
|
213 |
+
"params": {"k": 5, "max_length": 100},
|
214 |
+
},
|
215 |
}
|
216 |
|
217 |
# Define the order for display.
|
|
|
223 |
"Min-p Sampling",
|
224 |
"Eta Sampling",
|
225 |
"Epsilon Sampling",
|
226 |
+
"laconic",
|
227 |
+
"COT Decoding",
|
228 |
]
|
229 |
results = {method: None for method in methods}
|
230 |
|
|
|
243 |
future = executor.submit(
|
244 |
generate_min_p_completion, prompt, info["pbase"]
|
245 |
)
|
246 |
+
elif method == "laconic":
|
247 |
+
future = executor.submit(generate_laconic_completion, prompt)
|
248 |
+
elif method == "COT Decoding":
|
249 |
+
future = executor.submit(cot_decoding, prompt, **info["params"])
|
250 |
+
|
251 |
future_to_method[future] = method
|
252 |
|
253 |
# As each future completes, update its result and yield the current state.
|
|
|
275 |
gr.Textbox(label="Top-k Sampling"),
|
276 |
gr.Textbox(label="Top-p Sampling"),
|
277 |
gr.Textbox(label="Beam Search"),
|
278 |
+
gr.Textbox(label="Min-p Sampling (as in https://arxiv.org/abs/2407.01082)"),
|
279 |
gr.Textbox(label="Eta Sampling"),
|
280 |
gr.Textbox(label="Epsilon Sampling"),
|
281 |
+
gr.Textbox(
|
282 |
+
label="laconic decoding (by Alex Dimakis, 2025, search for twitter thread)"
|
283 |
+
),
|
284 |
+
gr.Textbox(
|
285 |
+
label="COT Decoding (Chain-of-Thought Reasoning without Prompting, Wang, Zhou, 2024)"
|
286 |
+
),
|
287 |
],
|
288 |
title="Decoding Methods Comparison",
|
289 |
description="Each decoding method's final answer is printed as soon as it is done. Model used: GPT-2.",
|