Support LLM api of the new trtllm (0.20.0) (#26)

* Use tensorrt llm api

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Update readme

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Improve GenLength LP example

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Update trtllm CiteFromPrompt LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Update trtllm LastPhrase LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Update trtllm LastPhrase LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Update trtllm MultipleChoice LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Add trtllm PreventHallucination LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Add trtllm TriggerPhrase LP

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Use pytorch backend

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

* Rename example_notebooks to examples

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

---------

Signed-off-by: aerdem4 <ahmeterd4@gmail.com>

Author: aerdem4

README.md

@@ -18,7 +18,7 @@ pip install logits-processor-zoo
 ## Supported Frameworks
 * transformers
 * vLLM
-* TensorRT-LLM
+* TensorRT-LLM (>=0.20.0)
 
 ## Usage
 
@@ -87,3 +87,5 @@ One common use case is to force writing python code just after thinking:
 trigger_python = TriggerPhraseLogitsProcessor(phrase="\n```python", trigger_token_phrase="</think>", 
                                               tokenizer=tokenizer, trigger_count=1, trigger_after=True)
 ```
+### PreventHallucinationLogitsProcessor
+A logits processor that mitigates hallucinated model outputs by enforcing a predefined fallback phrase when token confidence falls below a specified threshold.

example_notebooks/trtllm/README.md

@@ -33,7 +33,7 @@
     }
    ],
    "source": [
-    "from example_notebooks.transformers.utils import LLMRunner\n",
+    "from examples.transformers.utils import LLMRunner\n",
     "from logits_processor_zoo.transformers import CiteFromPromptLogitsProcessor\n",
     "\n",
     "\n",

example_notebooks/trtllm/cite_prompt_logits_processor.py

@@ -37,7 +37,7 @@
     }
    ],
    "source": [
-    "from example_notebooks.transformers.utils import LLMRunner\n",
+    "from examples.transformers.utils import LLMRunner\n",
     "from logits_processor_zoo.transformers import ForceLastPhraseLogitsProcessor\n",
     "\n",
     "\n",

example_notebooks/trtllm/gen_length_logits_processor.py

@@ -25,7 +25,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "from example_notebooks.transformers.utils import LLMRunner\n",
+    "from examples.transformers.utils import LLMRunner\n",
     "from logits_processor_zoo.transformers import GenLengthLogitsProcessor\n",
     "\n",
     "example_prompts =[\n",

example_notebooks/trtllm/multiple_choice_logits_processor.py

@@ -37,7 +37,7 @@
     }
    ],
    "source": [
-    "from example_notebooks.transformers.utils import LLMRunner\n",
+    "from examples.transformers.utils import LLMRunner\n",
     "from logits_processor_zoo.transformers import MultipleChoiceLogitsProcessor\n",
     "\n",
     "\n",