Merge branch 'dev'

Author: thiswillbeyourgithub

DocToolsLLM/DocToolsLLM.py

@@ -78,7 +78,7 @@
 class DocToolsLLM_class:
     "This docstring is dynamically replaced by the content of DocToolsLLM/docs/USAGE.md"
 
-    VERSION: str = "0.35"
+    VERSION: str = "0.36"
 
     #@optional_typecheck
     @typechecked
@@ -1282,7 +1282,7 @@ async def eval(inputs):
                         assert len(out.generations) == 1, f"Query eval llm produced more than 1 evaluations: '{out.generations}'"
                         outputs.append(out.generations[0].text)
                         finish_reason = out.generations[0].generation_info["finish_reason"]
-                        assert finish_reason == "stop", f"unexpected finish_reason: '{finish}'"
+                        assert finish_reason == "stop", f"unexpected finish_reason: '{finish_reason}'"
                         new_p += out.llm_output["token_usage"]["prompt_tokens"]
                         new_c += out.llm_output["token_usage"]["completion_tokens"]
                     assert outputs, "No generations found by query eval llm"

DocToolsLLM/utils/llm.py

@@ -84,7 +84,7 @@ def load_llm(
     if not private and backend == "openai" and api_base is None and llm_cache is not False:
         red("Using ChatOpenAI instead of litellm because calling openai server anyway and the caching has a bug on langchain side :( The caching works on ChatOpenAI though. More at https://github.com/langchain-ai/langchain/issues/22389")
         max_tokens = litellm.get_model_info(modelname)["max_tokens"]
-        if max_tokens not in extra_model_args:
+        if "max_tokens" not in extra_model_args:
             extra_model_args["max_tokens"] = max_tokens
         llm = ChatOpenAI(
                 model_name=modelname.split("/")[1],
@@ -96,7 +96,7 @@ def load_llm(
     else:
         red("A bug on langchain's side forces DocToolsLLM to disable the LLM caching. More at https://github.com/langchain-ai/langchain/issues/22389")
         max_tokens = litellm.get_model_info(modelname)["max_tokens"]
-        if max_tokens not in extra_model_args:
+        if "max_tokens" not in extra_model_args:
             extra_model_args["max_tokens"] = max_tokens
         if llm_cache is not False:
             red(f"Reminder: caching is disabled for non openai models until langchain approves the fix.")

DocToolsLLM/utils/misc.py

@@ -297,6 +297,7 @@ def get_splitter(
     max_tokens = 4096
     try:
         max_tokens = litellm.get_model_info(modelname)["max_input_tokens"]
+        # max_tokens = litellm.get_model_info(modelname)["max_tokens"]
 
         # don't use overly large chunks anyway
         max_tokens = min(max_tokens, int(2.5 * litellm.get_model_info(modelname)["max_tokens"]))

bumpver.toml

@@ -1,5 +1,5 @@
 [bumpver]
-current_version = "0.35"
+current_version = "0.36"
 version_pattern = "MAJOR.MINOR"
 commit_message = "bump version {old_version} -> {new_version}"
 tag_message = "{new_version}"

completion.cli.bash

@@ -95,167 +95,167 @@ _complete-DocToolsLLM()
   lastcommand=$(get_lastcommand)
 
   opts=""
-  GLOBAL_OPTIONS="--dollar-limit --no-llm-cache --query --top-k --load-embeds-from --llms-api-bases --embed-kwargs --query-eval-modelname --debug --task --query-retrievers --embed-model --chat-memory --summary-language --import-mode --save-embeds-as --query-eval-check-number --private --notification-callback --filetype --query-relevancy --query-condense-question --file-loader-parallel-backend --summary-n-recursion --modelname --DIY-rolling-window-embedding --llm-verbosity"
+  GLOBAL_OPTIONS="--debug --llm-verbosity --load-embeds-from --query-retrievers --notification-callback --query-eval-check-number --embed-model --DIY-rolling-window-embedding --llms-api-bases --chat-memory --filetype --query-condense-question --summary-n-recursion --query-eval-modelname --import-mode --save-embeds-as --query-relevancy --no-llm-cache --query --file-loader-parallel-backend --summary-language --embed-kwargs --task --dollar-limit --private --modelname --top-k"
 
 
   case "${lastcommand}" in
 
-    center)
+    translate)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--fillchar --width ${GLOBAL_OPTIONS}"
+        opts="--table ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    ljust)
+    removeprefix)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--fillchar --width ${GLOBAL_OPTIONS}"
+        opts="--prefix ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    prepare-query-task)
+    removesuffix)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--self ${GLOBAL_OPTIONS}"
+        opts="--suffix ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    expandtabs)
+    replace)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--tabsize ${GLOBAL_OPTIONS}"
+        opts="--count --new --old ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    encode)
+    split)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--encoding --errors ${GLOBAL_OPTIONS}"
+        opts="--maxsplit --sep ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    rstrip)
+    join)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--chars ${GLOBAL_OPTIONS}"
+        opts="--iterable ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    removesuffix)
+    VERSION)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--suffix ${GLOBAL_OPTIONS}"
+        opts="capitalize casefold center count encode endswith expandtabs find format format-map index isalnum isalpha isascii isdecimal isdigit isidentifier islower isnumeric isprintable isspace istitle isupper join ljust lower lstrip maketrans partition removeprefix removesuffix replace rfind rindex rjust rpartition rsplit rstrip split splitlines startswith strip swapcase title translate upper zfill ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    rpartition)
+    zfill)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--sep ${GLOBAL_OPTIONS}"
+        opts="--width ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    strip)
+    rpartition)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--chars ${GLOBAL_OPTIONS}"
+        opts="--sep ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    partition)
+    strip)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--sep ${GLOBAL_OPTIONS}"
+        opts="--chars ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    translate)
+    lstrip)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--table ${GLOBAL_OPTIONS}"
+        opts="--chars ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    lstrip)
+    ljust)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--chars ${GLOBAL_OPTIONS}"
+        opts="--fillchar --width ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    rjust)
+    expandtabs)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--fillchar --width ${GLOBAL_OPTIONS}"
+        opts="--tabsize ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    join)
+    rstrip)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--iterable ${GLOBAL_OPTIONS}"
+        opts="--chars ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    rsplit)
+    prepare-query-task)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--maxsplit --sep ${GLOBAL_OPTIONS}"
+        opts="--self ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    replace)
+    rsplit)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--count --new --old ${GLOBAL_OPTIONS}"
+        opts="--maxsplit --sep ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
@@ -266,52 +266,52 @@ _complete-DocToolsLLM()
       opts=$(filter_options $opts)
     ;;
 
-    split)
+    encode)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--maxsplit --sep ${GLOBAL_OPTIONS}"
+        opts="--encoding --errors ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    splitlines)
+    rjust)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--keepends ${GLOBAL_OPTIONS}"
+        opts="--fillchar --width ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    removeprefix)
+    partition)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--prefix ${GLOBAL_OPTIONS}"
+        opts="--sep ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    VERSION)
+    splitlines)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="capitalize casefold center count encode endswith expandtabs find format format-map index isalnum isalpha isascii isdecimal isdigit isidentifier islower isnumeric isprintable isspace istitle isupper join ljust lower lstrip maketrans partition removeprefix removesuffix replace rfind rindex rjust rpartition rsplit rstrip split splitlines startswith strip swapcase title translate upper zfill ${GLOBAL_OPTIONS}"
+        opts="--keepends ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;
 
-    zfill)
+    center)
 
       if is_prev_global; then
         opts="${GLOBAL_OPTIONS}"
       else
-        opts="--width ${GLOBAL_OPTIONS}"
+        opts="--fillchar --width ${GLOBAL_OPTIONS}"
       fi
       opts=$(filter_options $opts)
     ;;