Merge pull request #693 from ScrapeGraphAI/togheter_fixing

Togheter fixing
This commit is contained in:
Marco Vinciguerra 2024-09-24 09:36:59 +02:00 committed by GitHub
commit a557f13a09
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 18 additions and 5 deletions

View File

@ -154,12 +154,13 @@ class AbstractGraph(ABC):
try:
self.model_token = models_tokens[llm_params["model_provider"]][llm_params["model"]]
except KeyError:
print(f"""Model {llm_params['model_provider']}/{llm_params['model']} not found,
print(f"""Model {llm_params['model_provider']}/{llm_params['model']} not found,
using default token size (8192)""")
self.model_token = 8192
try:
if llm_params["model_provider"] not in {"oneapi","nvidia","ernie","deepseek","togetherai"}:
if llm_params["model_provider"] not in \
{"oneapi","nvidia","ernie","deepseek","togetherai"}:
if llm_params["model_provider"] == "bedrock":
llm_params["model_kwargs"] = { "temperature" : llm_params.pop("temperature") }
with warnings.catch_warnings():
@ -195,7 +196,7 @@ class AbstractGraph(ABC):
return ChatNVIDIA(**llm_params)
except Exception as e:
print(f"Error instancing model: {e}")
raise Exception(f"Error instancing model: {e}")
def get_state(self, key=None) -> dict:

View File

@ -131,9 +131,21 @@ models_tokens = {
"gemma-7b-it": 8192,
"claude-3-haiku-20240307'": 8192,
},
"togheterai": {
"toghetherai": {
"meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo": 128000,
"meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": 128000
"meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo": 128000,
"mistralai/Mixtral-8x22B-Instruct-v0.1": 128000,
"stabilityai/stable-diffusion-xl-base-1.0": 2048,
"meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo": 128000,
"NousResearch/Hermes-3-Llama-3.1-405B-Turbo": 128000,
"Gryphe/MythoMax-L2-13b-Lite": 8192,
"Salesforce/Llama-Rank-V1": 8192,
"meta-llama/Meta-Llama-Guard-3-8B": 128000,
"meta-llama/Meta-Llama-3-70B-Instruct-Turbo": 128000,
"meta-llama/Llama-3-8b-chat-hf": 8192,
"meta-llama/Llama-3-70b-chat-hf": 8192,
"Qwen/Qwen2-72B-Instruct": 128000,
"google/gemma-2-27b-it": 8192
},
"anthropic": {
"claude_instant": 100000,