lollms-client 0.13.1__py3-none-any.whl → 0.13.2__py3-none-any.whl

examples/simple_text_gen_test.py

@@ -1,4 +1,4 @@
-from lollms_client import LollmsClient, ELF_COMPLETION_FORMAT
+from lollms_client import LollmsClient
 from lollms_client.lollms_types import MSG_TYPE # For callback signature
 from ascii_colors import ASCIIColors, trace_exception
 

lollms_client/__init__.py

@@ -6,7 +6,7 @@ from lollms_client.lollms_discussion import LollmsDiscussion, LollmsMessage
 from lollms_client.lollms_utilities import PromptReshaper # Keep general utilities
 from lollms_client.lollms_functions import FunctionCalling_Library
 
-__version__ = "0.13.1"
+__version__ = "0.13.2"
 
 # Optionally, you could define __all__ if you want to be explicit about exports
 __all__ = [

lollms_client/llm_bindings/ollama/__init__.py

@@ -10,10 +10,11 @@ from typing import Optional, Callable, List, Union, Dict
 
 from ascii_colors import ASCIIColors, trace_exception
 import pipmaster as pm
-pm.ensure_packages(["ollama","pillow"])
+pm.ensure_packages(["ollama","pillow","tiktoken"])
 
 
 import ollama
+import tiktoken
 BindingName = "OllamaBinding"
 
 
@@ -236,53 +237,31 @@ class OllamaBinding(LollmsLLMBinding):
             trace_exception(ex)
             return {"status": False, "error": error_message}
     
-    def tokenize(self, text: str) -> List[Union[int, str]]:
+    def tokenize(self, text: str) -> list:
         """
-        Tokenize the input text. For Ollama, this is complex as tokenization is model-specific
-        and best done by the server. This method provides a basic character-level tokenization
-        as a fallback or placeholder, or one could attempt to call /api/tokenize if desired.
-        The `count_tokens` method is more accurate for Ollama.
+        Tokenize the input text into a list of characters.
 
         Args:
             text (str): The text to tokenize.
 
         Returns:
-            list: List of tokens (characters or token IDs if /api/tokenize is used).
+            list: List of individual characters.
         """
-        # Basic character-level tokenization
-        # return list(text)
-
-        # For actual token IDs (slower, makes a network request):
-        api_url = f"{self.host_address.rstrip('/')}/api/tokenize"
-        payload = {"model": self.model_name, "prompt": text}
-        try:
-            response = requests.post(api_url, json=payload, timeout=10, verify=self.verify_ssl_certificate, headers=self.ollama_client_headers)
-            response.raise_for_status()
-            return response.json().get("tokens", [])
-        except Exception as e:
-            ASCIIColors.warning(f"Failed to tokenize text with Ollama server, falling back to char tokens: {e}")
-            return list(text)
-
-    def detokenize(self, tokens: List[Union[int,str]]) -> str:
+        ## Since ollama has no endpoints to tokenize the text, we use tiktoken to have a rough estimate
+        return tiktoken.model.encoding_for_model("gpt-3.5-turbo").encode(text)
+            
+    def detokenize(self, tokens: list) -> str:
         """
-        Convert a list of tokens back to text. If tokens are characters, joins them.
-        If tokens are IDs, this is non-trivial without the model's tokenizer.
+        Convert a list of tokens back to text.
 
         Args:
-            tokens (list): List of tokens to detokenize.
+            tokens (list): List of tokens (characters) to detokenize.
 
         Returns:
             str: Detokenized text.
         """
-        if not tokens:
-            return ""
-        if isinstance(tokens[0], str): # Assuming character tokens
-            return "".join(tokens)
-        else: 
-            # Detokenizing IDs from Ollama is not straightforward client-side without specific tokenizer.
-            # This is a placeholder. For Ollama, detokenization usually happens server-side.
-            ASCIIColors.warning("Detokenizing integer tokens is not accurately supported by this Ollama client binding. Returning joined string of token IDs.")
-            return "".join(map(str, tokens))
+        ## Since ollama has no endpoints to tokenize the text, we use tiktoken to have a rough estimate
+        return tiktoken.model.encoding_for_model("gpt-3.5-turbo").decode(tokens)
     
     def count_tokens(self, text: str) -> int:
         """
@@ -297,8 +276,8 @@ class OllamaBinding(LollmsLLMBinding):
         if not self.model_name:
             ASCIIColors.warning("Cannot count tokens, model_name is not set.")
             return -1
-        return count_tokens_ollama(text, self.model_name, self.ollama_client)
-    
+        #return count_tokens_ollama(text, self.model_name, self.ollama_client)
+        return len(self.tokenize(text))
     def embed(self, text: str, **kwargs) -> List[float]:
         """
         Get embeddings for the input text using Ollama API.

{lollms_client-0.13.1.dist-info → lollms_client-0.13.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: lollms_client
-Version: 0.13.1
+Version: 0.13.2
 Summary: A client library for LoLLMs generate endpoint
 Author-email: ParisNeo <parisneoai@gmail.com>
 License: Apache Software License

{lollms_client-0.13.1.dist-info → lollms_client-0.13.2.dist-info}/RECORD

@@ -1,4 +1,4 @@
-examples/simple_text_gen_test.py,sha256=cwXpwe_9V_Vgmn9ibe-QH3lAudL1Jxw0KrzkcmYllI4,8733
+examples/simple_text_gen_test.py,sha256=RoX9ZKJjGMujeep60wh5WT_GoBn0O9YKJY6WOy-ZmOc,8710
 examples/simple_text_gen_with_image_test.py,sha256=Euv53jbKTVJDvs854lgJvA5F-iRnAATLxAklig24ots,8534
 examples/text_2_audio.py,sha256=MfL4AH_NNwl6m0I0ywl4BXRZJ0b9Y_9fRqDIe6O-Sbw,3523
 examples/text_2_image.py,sha256=Ri7lQ-GW54YWQh2eofcaN6LpwFoorbpJsJffrcXl3cg,6415
@@ -12,7 +12,7 @@ examples/personality_test/chat_test.py,sha256=o2jlpoddFc-T592iqAiA29xk3x27KsdK5D
 examples/personality_test/chat_with_aristotle.py,sha256=4X_fwubMpd0Eq2rCReS2bgVlUoAqJprjkLXk2Jz6pXU,1774
 examples/personality_test/tesks_test.py,sha256=7LIiwrEbva9WWZOLi34fsmCBN__RZbPpxoUOKA_AtYk,1924
 examples/test_local_models/local_chat.py,sha256=slakja2zaHOEAUsn2tn_VmI4kLx6luLBrPqAeaNsix8,456
-lollms_client/__init__.py,sha256=dYRkaz6dRRkKVNICLLMXkZsK_pDFPo7A1ukkGdv-aoU,823
+lollms_client/__init__.py,sha256=y-N8Dw10pI9pHtP_zlVzsj7bVjsu873EdOlbqxaxZRU,823
 lollms_client/lollms_config.py,sha256=goEseDwDxYJf3WkYJ4IrLXwg3Tfw73CXV2Avg45M_hE,21876
 lollms_client/lollms_core.py,sha256=ZTbEVn1M_gHAL3mL5mf3wGYAXidAtnSI3qEjwz2HlwY,77980
 lollms_client/lollms_discussion.py,sha256=9b83m0D894jwpgssWYTQHbVxp1gJoI-J947Ui_dRXII,2073
@@ -30,7 +30,7 @@ lollms_client/lollms_types.py,sha256=cfc1sremM8KR4avkYX99fIVkkdRvXErrCWKGjLrgv50
 lollms_client/lollms_utilities.py,sha256=YAgamfp0pBVApR68AHKjhp1lh6isMNF8iadwWLl63c0,7045
 lollms_client/llm_bindings/__init__.py,sha256=9sWGpmWSSj6KQ8H4lKGCjpLYwhnVdL_2N7gXCphPqh4,14
 lollms_client/llm_bindings/lollms/__init__.py,sha256=l1q2KnMQALz9QpLa3OUQ8e29KU4RCwkrmrdBvd7Z_kc,12236
-lollms_client/llm_bindings/ollama/__init__.py,sha256=8gxYzqun08fU5oXDWpbgXmAOuHeGxjx-dtpn_SqmqxI,27761
+lollms_client/llm_bindings/ollama/__init__.py,sha256=DyueED1cJmmJFg5evYmu-lrkwsN9pAxaVcwgUkcAZHU,26467
 lollms_client/llm_bindings/openai/__init__.py,sha256=SWBgnOcOWmFRSKTN1S9ATownHNBJ9f6FEtI3L4xNJNM,11861
 lollms_client/llm_bindings/tensor_rt/__init__.py,sha256=ZpeSKAbN8rh6zkysYl95sXG9Ci702NuPAhXC6zb1zT4,31840
 lollms_client/llm_bindings/transformers/__init__.py,sha256=8JbX3B-obLt5NNtcNOGD_E0f8OQTma2pNYtVt2urTOM,12572
@@ -45,8 +45,8 @@ lollms_client/tts_bindings/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJ
 lollms_client/tts_bindings/lollms/__init__.py,sha256=8x2_T9XscvISw2TiaLoFxvrS7TIsVLdqbwSc04cX-wc,7164
 lollms_client/ttv_bindings/__init__.py,sha256=UZ8o2izQOJLQgtZ1D1cXoNST7rzqW22rL2Vufc7ddRc,3141
 lollms_client/ttv_bindings/lollms/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-lollms_client-0.13.1.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-lollms_client-0.13.1.dist-info/METADATA,sha256=eLvEitmwnBztnm_z5Cb-aKh_FyW6DTMWqgxMvNcRwos,7276
-lollms_client-0.13.1.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
-lollms_client-0.13.1.dist-info/top_level.txt,sha256=NI_W8S4OYZvJjb0QWMZMSIpOrYzpqwPGYaklhyWKH2w,23
-lollms_client-0.13.1.dist-info/RECORD,,
+lollms_client-0.13.2.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+lollms_client-0.13.2.dist-info/METADATA,sha256=IsZiVKLRi7NaMhITn6pFz3CX5zpIJAvb8vsddYIBJQA,7276
+lollms_client-0.13.2.dist-info/WHEEL,sha256=Nw36Djuh_5VDukK0H78QzOX-_FQEo6V37m3nkm96gtU,91
+lollms_client-0.13.2.dist-info/top_level.txt,sha256=NI_W8S4OYZvJjb0QWMZMSIpOrYzpqwPGYaklhyWKH2w,23
+lollms_client-0.13.2.dist-info/RECORD,,