NgzddlZddlmZddlmZddlmZmZmZm Z m Z m Z m Z m Z mZmZmZddlmZddlmZddlmZmZddlmZmZmZmZmZmZmZm Z m!Z!m"Z"m#Z#m$Z$m%Z%m&Z&dd l'm(Z(m)Z)m*Z*dd l+m,Z,dd l-m.Z.m/Z/m0Z0m1Z1dd l2m3Z3m4Z4m5Z5dd l6m7Z7m8Z8m9Z9ddl:m;Z;ddlm?Z?ddl@mAZAmBZBmCZCddlDmEZEGddeZFde)deGfdZHde(deGfdZIde eJefdefdZKdedefdZLdedeGfdZMde eJefd eedefd!ZNdS)"N) itemgetter)Path) AnyCallableDictIteratorListMappingOptionalSequenceTypeUnioncast)CallbackManagerForLLMRun)LanguageModelInput) BaseChatModelgenerate_from_stream) AIMessageAIMessageChunk BaseMessageBaseMessageChunk ChatMessageChatMessageChunkFunctionMessageFunctionMessageChunk HumanMessageHumanMessageChunk SystemMessageSystemMessageChunk ToolMessageToolMessageChunk)InvalidToolCallToolCall ToolCallChunk)OutputParserLike)JsonOutputKeyToolsParserPydanticToolsParsermake_invalid_tool_callparse_tool_call)ChatGenerationChatGenerationChunk ChatResult)Runnable RunnableMapRunnablePassthrough)BaseToolconvert_to_openai_tool)is_basemodel_subclass) BaseModelFieldmodel_validator)Selfc~eZdZUdZdZeed<eed< dZe eed< dZ e eed< dZ e ed< d Z e ed < d Ze ed < d Zeed < dZeed< dZeed< dZeed< dZe e ed< dZe ed< dZe e ed< dZe eed< dZe ed< dZeed< dZeed< dZe e ed< dZeed< dZe eeed< d Z eed!< d"Z!e ed#< d$Z"e ed%< d Z#eed&< d'Z$eed(< d)Z%eed*< e&e'+Z(e)eefed,< d Z*eed-< dZ+e e,ee-fed.< dZ.eed/< d Z/eed0< e0d12d3e1fd4Z2de eed3e)eeffd5Z3d6ee4d3ee)eeffd7Z5d8e'd3e6fd9Z7 dId6ee4de eed:e e8d;ed3e6f d<Z9 dId6ee4de eed:e e8d;ed3e:e;f d=Zd?e=e,e)eefe>e?e@eAfd@e e,e'eefd;ed3eBeCe4fffdAZD dJddBdCe e,e)e>e?fdDed;ed3eBeCe,e)e?fffdEZEeFd3e)eeffdFZGeFd3efdGZHeFd3e)eeffdHZIxZJS)K ChatLlamaCppzllama.cpp model. To use, you should have the llama-cpp-python library installed, and provide the path to the Llama model as a named parameter to the constructor. Check out: https://github.com/abetlen/llama-cpp-python Nclient model_path lora_base lora_pathin_ctxn_partsseedTf16_kvF logits_all vocab_only use_mlock n_threadsn_batch n_gpu_layerssuffix max_tokensg? temperaturegffffff?top_plogprobsechostopg?repeat_penalty(top_k@last_n_tokens_sizeuse_mmapg?rope_freq_scaleg@rope_freq_base)default_factory model_kwargs streaming grammar_pathgrammarverboseafter)modereturnct ddlm}m}n#t$rtdwxYwj}gd}fd|D}j j|d<|j ||fi|_n'#t$r}td|d |d}~wwxYwj r+j r$j }j }td |d |d tj tr |j _ n(j r |j _ n S) z4Validate that llama-cpp-python library is installed.r)Llama LlamaGrammarzCould not import llama-cpp-python library. Please install the llama-cpp-python library to use this embedding model: pip install llama-cpp-python)rXrYr=r<r>r@rArBrCrDrErFrHrWrVr_c2i|]}|t|S)getattr).0kselfs d/var/www/html/ai-engine/env/lib/python3.11/site-packages/langchain_community/chat_models/llamacpp.py z5ChatLlamaCpp.validate_environment..s%GGG74++GGGNrIz&Could not load Llama model from path: z. Received error zCCan only pass in one of grammar and grammar_path. Received grammar=z and grammar_path=.) llama_cpprdre ImportErrorr;rIupdater[r: Exception ValueErrorr^r] isinstancestr from_string from_file) rkrdrer;model_param_names model_paramser^r]s ` rlvalidate_environmentz!ChatLlamaCpp.validate_environments  5 5 5 5 5 5 5 5 5   I  _    $HGGG5FGGG   (+/+z6ChatLlamaCpp._create_message_dicts..s!GGG1!44GGGrnrg)rkr message_dictss rl_create_message_dictsz"ChatLlamaCpp._create_message_dictssHGhGGG rnresponsecLg}|dD]o}t|d}t|d}d|vr |d|d<t||}||p|di}d|i}t || S) Nchoicesmessage finish_reason)rrOrgeneration_infousage token_usage) generations llm_output)_convert_dict_to_messagedictgetr*appendr,) rkrrresrrgenrrs rl_create_chat_resultz ChatLlamaCpp._create_chat_results I& $ $C.s9~>>G"1I1IJJJOS  .1*o + /RRRC   s # # # #ll7B// ; kjIIIIrn run_managerkwargsc &i|||}|jr4|ds|j|fd|i|}t |S||}|jjdd|i|}||S)N tool_choicerrrg) rr\r_streamrrr:create_chat_completionr) rkrrQrrr stream_iterrrs rl _generatezChatLlamaCpp._generates:D((..9&9 > 5&**]";"; 5&$,xSS[SFSSK' 44 4228<< 54;5WW}WPVWW''111rnc+Ki|||}||}|jjd |dd|}t}d} |D]} | dz } t | t s| } t| ddkrJ| dd} | dat| d|} i} | dx} r| | d<| d}|r|| d<| j }t| | pd } |r| | j| | | VdS) NT)rstreamrrdeltarrOr)chunkrOrg)rrr:rrrur model_dumplen_convert_delta_to_message_chunkr __class__r+on_llm_new_tokentext)rkrrQrrrrresultdefault_chunk_classcountrchoicerrrOs rlrzChatLlamaCpp._stream+s:D((..9&9228<< 33 "4  39  -  E QJEeT** +((**5#$$))9%a(Fg&3w!4E!O & ? ; ;;} A3@0zz*--H 7.6 +"'/ '/F$E Y,,UZux,XXXKKKK1  rnrtoolsrc ^d|D}d|D}rttr2tfd|Dstdd|dnttr'fd|D}|stdd|dnctt r>}}. c,g|]}t|Srgr1ritools rlrz+ChatLlamaCpp.bind_tools..b!JJJD1$77JJJrnc*g|]}|ddSfunctionnamerg)rifts rlrz+ChatLlamaCpp.bind_tools..cs!GGGbnV,GGGrnc3<K|]}dd|kVdS)rrNrg)rirrs rl z*ChatLlamaCpp.bind_tools..fsC@DK +F3t;rnzTool choice tool_choice=z1 was specified, but the only provided tools were roc8g|]}|ddk|Srrg)rifrs rlrz+ChatLlamaCpp.bind_tools..ns3!J-2G;2V2VA2V2V2VrnrzQtool_choice=True can only be specified when a single tool is passed in. Received z tools.rzUnrecognized tool_choice type. Expected dict having format like this {"type": "function", "function": {"name": <>}}Received: rc,g|]}t|Srgr1rs rlrz+ChatLlamaCpp.bind_tools..rrnrrg) ruranyrtrvboolrsuperbind)rkrrrformatted_tools tool_nameschosenrs ` rl bind_toolszChatLlamaCpp.bind_toolsUsKJEJJJGGGGG  +t,, HR%={==/9===K-- .$={==/9=== K.. ''!++$C/25zzCCC.a0  /!,// !,}JJEJJJuww|<", "<|eot_id|>"], ) structured_llm = llm.with_structured_output(AnswerWithJustification) structured_llm.invoke("What weighs more a pound of bricks or a pound of feathers") # -> AnswerWithJustification( # answer='They weigh the same', # justification='Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ.' # ) Example: Pydantic schema (include_raw=True): .. code-block:: python from langchain_community.chat_models import ChatLlamaCpp from pydantic import BaseModel class AnswerWithJustification(BaseModel): '''An answer to the user question along with justification for the answer.''' answer: str justification: str llm = ChatLlamaCpp( temperature=0., model_path="./SanctumAI-meta-llama-3-8b-instruct.Q8_0.gguf", n_ctx=10000, n_gpu_layers=4, n_batch=200, max_tokens=512, n_threads=multiprocessing.cpu_count() - 1, repeat_penalty=1.5, top_p=0.5, stop=["<|end_of_text|>", "<|eot_id|>"], ) structured_llm = llm.with_structured_output(AnswerWithJustification, include_raw=True) structured_llm.invoke("What weighs more a pound of bricks or a pound of feathers") # -> { # 'raw': AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_Ao02pnFYXD6GN1yzc0uXPsvF', 'function': {'arguments': '{"answer":"They weigh the same.","justification":"Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ."}', 'name': 'AnswerWithJustification'}, 'type': 'function'}]}), # 'parsed': AnswerWithJustification(answer='They weigh the same.', justification='Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ.'), # 'parsing_error': None # } Example: dict schema (include_raw=False): .. code-block:: python from langchain_community.chat_models import ChatLlamaCpp from pydantic import BaseModel from langchain_core.utils.function_calling import convert_to_openai_tool class AnswerWithJustification(BaseModel): '''An answer to the user question along with justification for the answer.''' answer: str justification: str dict_schema = convert_to_openai_tool(AnswerWithJustification) llm = ChatLlamaCpp( temperature=0., model_path="./SanctumAI-meta-llama-3-8b-instruct.Q8_0.gguf", n_ctx=10000, n_gpu_layers=4, n_batch=200, max_tokens=512, n_threads=multiprocessing.cpu_count() - 1, repeat_penalty=1.5, top_p=0.5, stop=["<|end_of_text|>", "<|eot_id|>"], ) structured_llm = llm.with_structured_output(dict_schema) structured_llm.invoke("What weighs more a pound of bricks or a pound of feathers") # -> { # 'answer': 'They weigh the same', # 'justification': 'Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume and density of the two substances differ.' # } zReceived unsupported arguments NzJschema must be specified when method is 'function_calling'. Received None.rr)typerrT)rfirst_tool_only)key_namerrawcdSNrg_s rlz5ChatLlamaCpp.with_structured_output..*sRVrn)parsed parsing_errorcdSrrgrs rlrz5ChatLlamaCpp.with_structured_output..,sdrn)rr) exception_key)r)rtrurr3r2rr'rr r&r/assignrwith_fallbacksr.) rkrrris_pydantic_schema tool_namerllm output_parser parser_assign parser_noneparser_with_fallbacks rlwith_structured_outputz#ChatLlamaCpp.with_structured_outputsX  IGvGGHH H'55W:OPV:W:W >! +622:>vF ) 7JKK oovhKo@@  .AD&))*D///MM5"DM  '/6!%((=8M.4NNKKKK#0#?#? _$@$$ 3'''*>> >& &rnc&id|ji|jS)zReturn a dictionary of identifying parameters. This information is used by the LangChain callback system, which is used for tracing purposes make it possible to monitor LLMs. r;)r;rrks rl_identifying_paramsz ChatLlamaCpp._identifying_params4s* T_-  "   rncdS)z7Get the type of language model used by this chat model.zllama-cpp-pythonrgrs rl _llm_typezChatLlamaCpp._llm_typeDs "!rnc|j|j|j|j|j|j|jd}|jr |j|d<|S)z>Get the default parameters for calling create_chat_completion.)rLrMrNrTrOr~rRr^)rLrMrNrTrOrQrRr^)rkrs rlrzChatLlamaCpp._default_paramsIsR/+ZZ "i"1   < - $ F9  rn)NNr)K__name__ __module__ __qualname____doc__r:r__annotations__rvr<r r=r>intr@rArBrrCrDrErFrHrIrJrLrMfloatrNrOrPrQr rRrTrVrWrXrYr5rr[rr\r]rrr^r_r6r7r|rrrr,rrrrr+rr r r4rr0r-rrrpropertyrrr __classcell__)rs@rlr9r9=sFCOOO+#Ix}###0#Ix}###AE3GS?D#NNN-FD1J@J/It,#Ix}###CGS/#'L(3-&&&F FHSM   SJ3K.E5."Hhsm"""ND$% $D(49 $$$@NE2E3OOO.    MHd1 OU   )#NE###+#(5#>#>#>L$sCx.>>>?It8/3L(5d+,333 GS GT)_'""":d:::#":xHT#Y$7DcN$[) d38n JDJZJJJJ&%):> 22{#2tCy!267 2  2  22220%):> (({#(tCy!(67 (  ( % & ((((\9= 1=1=1=d38nd9oxQRS1=eD$O45 1=  1= $k1 2 1=1=1=1=1=1=j:>j'" j'j'j'tT)_456j' j'  j' $eD)O&<< = j'j'j'j'X  T#s(^    X  "3"""X" c3h   X     rnr9 tool_callrbc\d|d|dtj|dddSNridrargs)r arguments)rrr)jsondumps)rs rl!_lc_tool_call_to_openai_tool_callrZs?of%If$566    rninvalid_tool_callc8d|d|d|dddSrrg)rs rl)_lc_invalid_tool_call_to_openai_tool_callres8%%f-*62    rn_dictc |d}|d}|d}|dkr&t|dd||S|dkr|ddpd}i}|d x}rt||d <g}g}|d x} r}| |d <| D]u} t| d } | s|| -#t $r<} t | t| } || Yd } ~ nd } ~ wwxYwt||||||S|dkr&t|dd||S|dkrLt|ddtt|d|S|dkr_i}d|vr |d|d<t|ddtt|d|||St|ddtt||S)zConvert a dictionary to a LangChain message. Args: _dict: The dictionary. Returns: The LangChain message. rolerrusercontent)rrr assistant function_call tool_callsT) return_idN)radditional_kwargsrrr invalid_tool_callssystemrrrrr tool_call_id)rrr rrrrr)rrrr)rrsr(rvrrrrr r)rrrid_rr r r r raw_tool_calls raw_tool_calltcr{ invalid_tcs rlrrrs 99V  D 99V  D ))D//C v~~EIIi$<$<4PPPP   ))Ir**0b"$!IIo66 6= E15m1D1D o . "YY|44 4> ..< l +!/ . .  .($GGGB . "))"----!:::!7 s1vv!N!NJ&--j99999999:/!1        UYYy"%=%=DSQQQQ   IIi,,4UYYv=N=N3O3OTW      U??(-f f %IIi,,c599^#<#<==/     IIi,,4T??s    sD E 2EE rc|rXt|trCg}|D]=}t|trd|vr |ddkr(||>n|}|S)zFormat message content.rtool_use)rulistrr)rformatted_contentblocks rl_format_message_contentrs $:gt,, $ 0 0E5$'' 0eOO&MZ//!((//// 0$ rnrcdt|ji}|jp|jdx}||d<t |t r |j|d<nat |trd|d<nEt |trd|d<d|jvr|jd|d<|j s|j r'd|j Dd |j Dz|d <n6d |jvr,|jd |d <hd fd |d D|d <n d|vsd |vr |dpd|d<nt |trd |d<nvt |trd|d<n[t |tr4d|d<|j|d<hdfd|D}nt#d||S)zConvert a LangChain message to a dictionary. Args: message: The LangChain message. Returns: The dictionary. rrNrrrr c,g|]}t|Srg)rrirs rlrz,_convert_message_to_dict..s.***:<1"55***rnc,g|]}t|Srg)rrs rlrz,_convert_message_to_dict..s0:"==rnr >rrrcPg|]"}fd|D#S)c$i|] \}}|v || Srgrg)rirjvtool_call_supported_propss rlrmz7_convert_message_to_dict...s*VVV$!Qq.sH***WVVV)//"3"3VVV***rnrrrr>rrrc$i|] \}}|v || Srgrg)rirjr#supported_propss rlrmz,_convert_message_to_dict..s)VVVAoAUAU1AUAUAUrnzGot unknown type )rrrr rrurrrrr r rrr rr% TypeError)r message_dictrr'r$s @@rlrrst *7?;;$L E 9 = =f E EER# V';''&7&| V G\ * *$7% V GY ' '"7* V g7 7 7,3,Eo,VL )   !; **@G@R***!4*L & & W6 6 6)0)B<)PL &(B(B(B %****!-l!;***L & &  l * *ll.J.J&29&=&EL # G] + + 7' V G_ - - 7) V G[ ) )7% V'.'; ^$===VVVV););)=)=VVV 5G55666 rn default_classcn|d}tt|d}tt|dpd}i}|dr+t|d}d|vr |dd|d<||d<g}|dx}r||d<|D]} t | dd| dd | d| d  } || w#t $rYwxYw|d ks |tkrt|| S|dks |tkrt||||S|dks |tkrt|| S|dks |tkrt||d|S|dks |tkrt||d|S|s |tkrt|||S||| S)Nrrrrr rr rrindex)rrrr,r)rrr)rr rtool_call_chunksrrrr)rrrr) rrrvrr$rKeyErrorrrrrr!r) rr*rrrr r r-rrtcrs rlrrs ))D//C UYYv&& ' 'D3 ),,233G  yy!!;U?344 ] " "}V'<'D$&M& !-:/*<000~ *8,'!  C )Z,,V44Z,,[99wwt}}g,  !'' 2222     v~~*;;; S9999     ? ?/-       ].@@@!'c::::   }0DDD#G%-CPPPP =,<<<%*?C     6"222dsCCCC}W5555s A3E E E)Oroperatorrpathlibrtypingrrrrr r r r r rrlangchain_core.callbacksrlangchain_core.language_modelsr*langchain_core.language_models.chat_modelsrrlangchain_core.messagesrrrrrrrrrrrrr r!langchain_core.messages.toolr"r#r$"langchain_core.output_parsers.baser%*langchain_core.output_parsers.openai_toolsr&r'r(r)langchain_core.outputsr*r+r,langchain_core.runnablesr-r.r/langchain_core.toolsr0%langchain_core.utils.function_callingr2langchain_core.utils.pydanticr3pydanticr4r5r6typing_extensionsr7r9rrrrvrrrrrgrnrlrAs                           >=========== RQQQQQQQQQ?????? SRRRRRRRRROOOOOOOOOO))))))HHHHHH?????? #"""""ZZZZZ=ZZZzd &      @ GCH$5@ +@ @ @ @ FSS(7k7d7777t/6 38 /6-12B-C/6/6/6/6/6/6/6rn