Nicholai/jan
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update ctxlen

Browse Source
This commit is contained in:
hahuyhoang411 2023-12-04 12:34:10 +07:00
parent dbc12d27c8
commit 88e37e1f2d
12 changed files with 23 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
models/neural-chat-7b/model.json
View File

@ -7,13 +7,13 @@
"description": "The Neural Chat 7B model, developed on the foundation of mistralai/Mistral-7B-v0.1, has been fine-tuned using the Open-Orca/SlimOrca dataset and aligned with the Direct Preference Optimization (DPO) algorithm. It has demonstrated substantial improvements in various AI tasks and performance well on the open_llm_leaderboard.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "### System: ",
"user_prompt": "### User: ",
"ai_prompt": "### Assistant: "
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "Intel, The Bloke",

4
models/neuralhermes-7b/model.json
View File

@ -7,13 +7,13 @@
"description": "NeuralHermes 2.5 has been enhanced using Direct Preference Optimization. This fine-tuning, inspired by the RLHF process of Neural-chat-7b and OpenHermes-2.5-Mistral-7B, has led to improved performance across several benchmarks.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "<|im_start|>system\n",
"user_prompt": "<|im_end|>\n<|im_start|>user\n",
"ai_prompt": "<|im_end|>\n<|im_start|>assistant\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "Intel, The Bloke",

4
models/noromaid-20b/model.json
View File

@ -7,13 +7,13 @@
"description": "The Noromaid 20b model is designed for role-playing and general use, featuring a unique touch with the no_robots dataset that enhances human-like behavior.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "",
"user_prompt": "### Instruction: ",
"ai_prompt": "### Response: "
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "NeverSleep, The Bloke",

4
models/openorca-13b/model.json
View File

@ -7,13 +7,13 @@
"description": "Orca 2 is a finetuned version of LLAMA-2, designed primarily for single-turn responses in reasoning, reading comprehension, math problem solving, and text summarization.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "<|im_start|>system\n",
"user_prompt": "<|im_end|>\n<|im_start|>user\n",
"ai_prompt": "<|im_end|>\n<|im_start|>assistant\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "Microsoft, The Bloke",

4
models/phind-34b/model.json
View File

@ -7,13 +7,13 @@
"description": "Phind-CodeLlama-34B-v2 is an AI model fine-tuned on 1.5B tokens of high-quality programming data. It's a SOTA open-source model in coding. This multi-lingual model excels in various programming languages, including Python, C/C++, TypeScript, Java, and is designed to be steerable and user-friendly.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "### System Prompt\n",
"user_prompt": "### User Message\n",
"ai_prompt": "### Assistant\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "Phind, The Bloke",

2
models/rocket-3b/model.json
View File

@ -13,7 +13,7 @@
"ai_prompt": "<|im_end|>\n<|im_start|>assistant\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "pansophic, The Bloke",

4
models/starling-7b/model.json
View File

@ -7,13 +7,13 @@
"description": "Starling-RM-7B-alpha is a language model finetuned with Reinforcement Learning from AI Feedback from Openchat 3.5. It stands out for its impressive performance using GPT-4 as a judge, making it one of the top-performing models in its category.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "",
"user_prompt": "GPT4 User: ",
"ai_prompt": "<|end_of_turn|>\nGPT4 Assistant: "
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "Berkeley-nest, The Bloke",

4
models/tiefighter-13b/model.json
View File

@ -7,13 +7,13 @@
"description": "Tiefighter-13B is a highly creative, merged AI model achieved by combining various 'LORAs' on top of an existing merge, particularly focusing on storytelling and improvisation. This model excels in story writing, chatbots, and adventuring, and is designed to perform better with less detailed inputs, leveraging its inherent creativity.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "",
"user_prompt": "### Instruction: ",
"ai_prompt": "\n### Response: "
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "KoboldAI, The Bloke",

4
models/wizardcoder-13b/model.json
View File

@ -7,13 +7,13 @@
"description": "WizardCoder-Python-13B is a Python coding model major models like ChatGPT-3.5. This model based on the Llama2 architecture, demonstrate high proficiency in specific domains like coding and mathematics.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "",
"user_prompt": "### Instruction:\n",
"ai_prompt": "### Response:\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "WizardLM, The Bloke",

4
models/wizardcoder-34b/model.json
View File

@ -7,13 +7,13 @@
"description": "WizardCoder-Python-34B is a Python coding model major models like ChatGPT-3.5. This model based on the Llama2 architecture, demonstrate high proficiency in specific domains like coding and mathematics.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "",
"user_prompt": "### Instruction:\n",
"ai_prompt": "### Response:\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "WizardLM, The Bloke",

4
models/yi-34b/model.json
View File

@ -7,13 +7,13 @@
"description": "Yi-34B, a specialized chat model, is known for its diverse and creative responses and excels across various NLP tasks and benchmarks.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "<|im_start|>system\n",
"user_prompt": "<|im_end|>\n<|im_start|>user\n",
"ai_prompt": "<|im_end|>\n<|im_start|>assistant\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "01-ai, The Bloke",

4
models/zephyr-beta-7b/model.json
View File

@ -7,13 +7,13 @@
"description": "The Zephyr-7B-β model marks the second iteration in the Zephyr series, designed to function as an effective assistant. It has been fine-tuned from the mistralai/Mistral-7B-v0.1 base model, utilizing a combination of public and synthetic datasets with the application of Direct Preference Optimization.",
"format": "gguf",
"settings": {
"ctx_len": 4096,
"ctx_len": 2048,
"system_prompt": "<|system|>\n",
"user_prompt": "</s>\n<|user|>\n",
"ai_prompt": "</s>\n<|assistant|>\n"
},
"parameters": {
"max_tokens": 4096
"max_tokens": 2048
},
"metadata": {
"author": "HuggingFaceH4, The Bloke",