From e6812b1247534c223d9e277353bec49fbc1785e5 Mon Sep 17 00:00:00 2001 From: Hoang Ha <64120343+hahuyhoang411@users.noreply.github.com> Date: Fri, 1 Dec 2023 17:20:58 +0700 Subject: [PATCH 1/2] chore: pre-populate Jan's /models folder with model.jsons (#775) * draft model.json * islm3b update * capybara 34b update * deepseek coder update * dolphin yi update * fix the maxtokens of islm * lzlv 70b update * marx3b update * mythomax 13b update * update neural chat 7b * noromaid 20b update * update openchat 7b * openhermes7b update * openorca 7b * orca 13b update * phind 34b update * rocket 3b update * starling 7b update * storytelling 70b update * tiefighter 13B * update tiefighter tags * tinyllama update * wizard coder 13b * update wizard coder 13b description * wizard coder 34b update * wizard coder minor fix * xwin 70b update * yarn 70b * yi 34b * zephyr beta 7b * neuralhermes-7b update * change path + ctxlen * update id * fix startling --- .gitignore | 1 - models/capybara-34b/model.json | 24 ++++++++++++++++++++++++ models/deepseek-coder-1.3b/model.json | 23 +++++++++++++++++++++++ models/dolphin-yi-34b/model.json | 24 ++++++++++++++++++++++++ models/islm-3b/model.json | 24 ++++++++++++++++++++++++ models/lzlv-70b/model.json | 24 ++++++++++++++++++++++++ models/marx-3b/model.json | 23 +++++++++++++++++++++++ models/mythomax-13b/model.json | 24 ++++++++++++++++++++++++ models/neural-chat-7b/model.json | 24 ++++++++++++++++++++++++ models/neuralhermes-7b/model.json | 24 ++++++++++++++++++++++++ models/noromaid-20b/model.json | 24 ++++++++++++++++++++++++ models/openchat-7b/model.json | 24 ++++++++++++++++++++++++ models/openhermes-mistral-7b/model.json | 24 ++++++++++++++++++++++++ models/openorca-13b/model.json | 24 ++++++++++++++++++++++++ models/openorca-7b/model.json | 24 ++++++++++++++++++++++++ models/phind-34b/model.json | 24 ++++++++++++++++++++++++ models/rocket-3b/model.json | 23 +++++++++++++++++++++++ models/starling-7b/model.json | 24 ++++++++++++++++++++++++ models/storytelling-70b/model.json | 24 ++++++++++++++++++++++++ models/tiefighter-13b/model.json | 24 ++++++++++++++++++++++++ models/tinyllama-1.1b/model.json | 23 +++++++++++++++++++++++ models/wizardcoder-13b/model.json | 24 ++++++++++++++++++++++++ models/wizardcoder-34b/model.json | 24 ++++++++++++++++++++++++ models/xwin-70b/model.json | 24 ++++++++++++++++++++++++ models/yarn-70b/model.json | 21 +++++++++++++++++++++ models/yi-34b/model.json | 24 ++++++++++++++++++++++++ models/zephyr-beta-7b/model.json | 24 ++++++++++++++++++++++++ 27 files changed, 617 insertions(+), 1 deletion(-) create mode 100644 models/capybara-34b/model.json create mode 100644 models/deepseek-coder-1.3b/model.json create mode 100644 models/dolphin-yi-34b/model.json create mode 100644 models/islm-3b/model.json create mode 100644 models/lzlv-70b/model.json create mode 100644 models/marx-3b/model.json create mode 100644 models/mythomax-13b/model.json create mode 100644 models/neural-chat-7b/model.json create mode 100644 models/neuralhermes-7b/model.json create mode 100644 models/noromaid-20b/model.json create mode 100644 models/openchat-7b/model.json create mode 100644 models/openhermes-mistral-7b/model.json create mode 100644 models/openorca-13b/model.json create mode 100644 models/openorca-7b/model.json create mode 100644 models/phind-34b/model.json create mode 100644 models/rocket-3b/model.json create mode 100644 models/starling-7b/model.json create mode 100644 models/storytelling-70b/model.json create mode 100644 models/tiefighter-13b/model.json create mode 100644 models/tinyllama-1.1b/model.json create mode 100644 models/wizardcoder-13b/model.json create mode 100644 models/wizardcoder-34b/model.json create mode 100644 models/xwin-70b/model.json create mode 100644 models/yarn-70b/model.json create mode 100644 models/yi-34b/model.json create mode 100644 models/zephyr-beta-7b/model.json diff --git a/.gitignore b/.gitignore index 9dfd1887c..a273404c2 100644 --- a/.gitignore +++ b/.gitignore @@ -2,7 +2,6 @@ .env # Jan inference -models/** error.log node_modules *.tgz diff --git a/models/capybara-34b/model.json b/models/capybara-34b/model.json new file mode 100644 index 000000000..562bcbe93 --- /dev/null +++ b/models/capybara-34b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Nous-Capybara-34B-GGUF/resolve/main/nous-capybara-34b.Q5_K_M.gguf", + "id": "capybara-34b", + "object": "model", + "name": "Capybara 200k 34B", + "version": 1.0, + "description": "Nous Capybara 34B, a variant of the Yi-34B model, is the first Nous model with a 200K context length, trained for three epochs on the innovative Capybara dataset.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "USER: ", + "ai_prompt": "ASSISTANT: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "NousResearch, The Bloke", + "tags": ["General", "Big Context Length"], + "size": 24320000000 + } + } + \ No newline at end of file diff --git a/models/deepseek-coder-1.3b/model.json b/models/deepseek-coder-1.3b/model.json new file mode 100644 index 000000000..2ff6d6e7b --- /dev/null +++ b/models/deepseek-coder-1.3b/model.json @@ -0,0 +1,23 @@ +{ + "source_url": "https://huggingface.co/TheBloke/deepseek-coder-1.3b-base-GGUF/resolve/main/deepseek-coder-1.3b-base.Q4_K_M.gguf", + "id": "deepseek-coder-1.3b", + "object": "model", + "name": "Deepseek Coder 1.3B", + "version": "1.0", + "description": "", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "", + "ai_prompt": "" + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "deepseek, The Bloke", + "tags": ["Code"], + "size": 870000000 + } + } diff --git a/models/dolphin-yi-34b/model.json b/models/dolphin-yi-34b/model.json new file mode 100644 index 000000000..3b1bf3619 --- /dev/null +++ b/models/dolphin-yi-34b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/dolphin-2_2-yi-34b-GGUF/resolve/main/dolphin-2_2-yi-34b.Q5_K_M.gguf", + "id": "dolphin-yi-34b", + "object": "model", + "name": "Dolphin Yi 34B", + "version": "1.0", + "description": "Dolphin, based on the Yi-34B model and enhanced with features like conversation and empathy, is trained on a unique dataset for advanced multi-turn conversations. Notably uncensored, it requires careful implementation of an alignment layer for ethical use.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "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 + }, + "metadata": { + "author": "ehartford, The Bloke", + "tags": ["General Use", "Role-playing"], + "size": 24320000000 + } + } + \ No newline at end of file diff --git a/models/islm-3b/model.json b/models/islm-3b/model.json new file mode 100644 index 000000000..916d7c50e --- /dev/null +++ b/models/islm-3b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/UmbrellaCorp/IS-LM-3B_GGUF/resolve/main/IS-LM-Q4_K_M.gguf", + "id": "islm-3b", + "object": "model", + "name": "IS LM 3B", + "version": "1.0", + "description": "IS LM 3B, based on the StableLM 3B model is specifically finetuned for economic analysis using DataForge Economics and QLoRA over three epochs, enhancing its proficiency in economic forecasting and analysis.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "USER: ", + "ai_prompt": "ASSISTANT: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "UmbrellaCorp, The Bloke", + "tags": ["General Use", "Economics"], + "size": 1710000000 + } + } + \ No newline at end of file diff --git a/models/lzlv-70b/model.json b/models/lzlv-70b/model.json new file mode 100644 index 000000000..4aaee79b3 --- /dev/null +++ b/models/lzlv-70b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/lzlv_70B-GGUF/resolve/main/lzlv_70b_fp16_hf.Q5_K_M.gguf", + "id": "lzlv-70b", + "object": "model", + "name": "Lzlv 70B", + "version": "1.0", + "description": "lzlv_70B is a sophisticated AI model designed for roleplaying and creative tasks. This merge aims to combine intelligence with creativity, seemingly outperforming its individual components in complex scenarios and creative outputs.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "USER: ", + "ai_prompt": "ASSISTANT: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "lizpreciatior, The Bloke", + "tags": ["General Use", "Role-playing"], + "size": 48750000000 + } + } + \ No newline at end of file diff --git a/models/marx-3b/model.json b/models/marx-3b/model.json new file mode 100644 index 000000000..78617d5c3 --- /dev/null +++ b/models/marx-3b/model.json @@ -0,0 +1,23 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Marx-3B-v3-GGUF/resolve/main/marx-3b-v3.Q4_K_M.gguf", + "id": "marx-3b", + "object": "model", + "name": "Marx 3B", + "version": "1.0", + "description": "Marx 3B, based on the StableLM 3B model is specifically finetuned for chating using EverythingLM data and QLoRA over two epochs, enhancing its proficiency in understand general knowledege.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "### HUMAN: ", + "ai_prompt": "### RESPONSE: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Bohan Du, The Bloke", + "tags": ["General Use"], + "size": 1620000000 + } + } \ No newline at end of file diff --git a/models/mythomax-13b/model.json b/models/mythomax-13b/model.json new file mode 100644 index 000000000..455f73968 --- /dev/null +++ b/models/mythomax-13b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/MythoMax-L2-13B-GGUF/resolve/main/mythomax-l2-13b.Q5_K_M.gguf", + "id": "mythomax-13b", + "object": "model", + "name": "Mythomax L2 13B", + "version": "1.0", + "description": "Mythomax L2 13b, an advanced AI model derived from MythoMix, merges MythoLogic-L2's deep comprehension with Huginn's writing skills through a unique tensor merge technique, excelling in roleplaying and storytelling.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "### Instruction: ", + "ai_prompt": "### Response: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Gryphe, The Bloke", + "tags": ["Role-playing"], + "size": 9230000000 + } + } + \ No newline at end of file diff --git a/models/neural-chat-7b/model.json b/models/neural-chat-7b/model.json new file mode 100644 index 000000000..f4f4f14d4 --- /dev/null +++ b/models/neural-chat-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/neural-chat-7B-v3-1-GGUF/resolve/main/neural-chat-7b-v3-1.Q4_K_M.gguf", + "id": "neural-chat-7b", + "object": "model", + "name": "Neural Chat 7B", + "version": "1.0", + "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, + "system_prompt": "### System: ", + "user_prompt": "### User: ", + "ai_prompt": "### Assistant: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Intel, The Bloke", + "tags": ["General Use", "Role-playing", "Big Context Length"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/neuralhermes-7b/model.json b/models/neuralhermes-7b/model.json new file mode 100644 index 000000000..07cca58d4 --- /dev/null +++ b/models/neuralhermes-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/NeuralHermes-2.5-Mistral-7B-GGUF/resolve/main/neuralhermes-2.5-mistral-7b.Q4_K_M.gguf", + "id": "neuralhermes-7b", + "object": "model", + "name": "NeuralHermes 7B", + "version": "1.0", + "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, + "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 + }, + "metadata": { + "author": "Intel, The Bloke", + "tags": ["General Use", "Code", "Big Context Length"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/noromaid-20b/model.json b/models/noromaid-20b/model.json new file mode 100644 index 000000000..86291e4f5 --- /dev/null +++ b/models/noromaid-20b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Noromaid-20B-v0.1.1-GGUF/resolve/main/noromaid-20b-v0.1.1.Q4_K_M.gguf", + "id": "noromaid-20b", + "object": "model", + "name": "Noromaid 20B", + "version": "1.0", + "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, + "system_prompt": "", + "user_prompt": "### Instruction: ", + "ai_prompt": "### Response: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "NeverSleep, The Bloke", + "tags": ["Role-playing"], + "size": 12040000000 + } + } + \ No newline at end of file diff --git a/models/openchat-7b/model.json b/models/openchat-7b/model.json new file mode 100644 index 000000000..1fd6bb259 --- /dev/null +++ b/models/openchat-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/openchat_3.5-GGUF/resolve/main/openchat_3.5.Q4_K_M.gguf", + "id": "openchat-7b", + "object": "model", + "name": "Open Chat 3.5 7B", + "version": "1.0", + "description": "OpenChat represents a breakthrough in the realm of open-source language models. By implementing the C-RLFT fine-tuning strategy, inspired by offline reinforcement learning, this 7B model achieves results on par with ChatGPT (March).", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "GPT4 User: ", + "ai_prompt": "<|end_of_turn|>\nGPT4 Assistant: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "OpenChat, The Bloke", + "tags": ["General", "Code"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/openhermes-mistral-7b/model.json b/models/openhermes-mistral-7b/model.json new file mode 100644 index 000000000..6b64363d5 --- /dev/null +++ b/models/openhermes-mistral-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/OpenHermes-2.5-Mistral-7B-GGUF/resolve/main/openhermes-2.5-mistral-7b.Q4_K_M.gguf", + "id": "openhermes-mistral-7b", + "object": "model", + "name": "Openhermes 2.5 Mistral 7B", + "version": "1.0", + "description": "The OpenHermes 2.5 Mistral 7B incorporates additional code datasets, more than a million GPT-4 generated data examples, and other high-quality open datasets. This enhancement led to significant improvement in benchmarks, highlighting its improved skill in handling code-centric tasks.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "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 + }, + "metadata": { + "author": "Teknium, The Bloke", + "tags": ["General", "Roleplay"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/openorca-13b/model.json b/models/openorca-13b/model.json new file mode 100644 index 000000000..02a555430 --- /dev/null +++ b/models/openorca-13b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Orca-2-13B-GGUF/resolve/main/orca-2-13b.Q5_K_M.gguf", + "id": "openorca-13b", + "object": "model", + "name": "Orca 2 13B", + "version": "1.0", + "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, + "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 + }, + "metadata": { + "author": "Microsoft, The Bloke", + "tags": ["General Use"], + "size": 9230000000 + } + } + \ No newline at end of file diff --git a/models/openorca-7b/model.json b/models/openorca-7b/model.json new file mode 100644 index 000000000..42c88212c --- /dev/null +++ b/models/openorca-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Mistral-7B-OpenOrca-GGUF/resolve/main/mistral-7b-openorca.Q4_K_M.gguf", + "id": "openorca-7b", + "object": "model", + "name": "OpenOrca 7B", + "version": "1.0", + "description": "OpenOrca 8k 7B is a model based on Mistral 7B, fine-tuned using the OpenOrca dataset. Notably ranked first on the HF Leaderboard for models under 30B, it excels in efficiency and accessibility.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "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 + }, + "metadata": { + "author": "OpenOrca, The Bloke", + "tags": ["General", "Code"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/phind-34b/model.json b/models/phind-34b/model.json new file mode 100644 index 000000000..4391ae08d --- /dev/null +++ b/models/phind-34b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Phind-CodeLlama-34B-v2-GGUF/resolve/main/phind-codellama-34b-v2.Q5_K_M.gguf", + "id": "phind-34b", + "object": "model", + "name": "Phind 34B", + "version": "1.0", + "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, + "system_prompt": "### System Prompt\n", + "user_prompt": "### User Message\n", + "ai_prompt": "### Assistant\n" + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Phind, The Bloke", + "tags": ["Code", "Big Context Length"], + "size": 24320000000 + } + } + \ No newline at end of file diff --git a/models/rocket-3b/model.json b/models/rocket-3b/model.json new file mode 100644 index 000000000..b00eb1f44 --- /dev/null +++ b/models/rocket-3b/model.json @@ -0,0 +1,23 @@ +{ + "source_url": "https://huggingface.co/TheBloke/rocket-3B-GGUF/resolve/main/rocket-3b.Q4_K_M.gguf", + "id": "rocket-3b", + "object": "model", + "name": "Rocket 3B", + "version": "1.0", + "description": "Rocket-3B is a GPT-like model, primarily English, fine-tuned on diverse public datasets. It outperforms larger models in benchmarks, showcasing superior understanding and text generation, making it an effective chat model for its size.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "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 + }, + "metadata": { + "author": "pansophic, The Bloke", + "tags": ["General Use"], + "size": 1710000000 + } + } \ No newline at end of file diff --git a/models/starling-7b/model.json b/models/starling-7b/model.json new file mode 100644 index 000000000..c029ea7d9 --- /dev/null +++ b/models/starling-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Starling-LM-7B-alpha-GGUF/resolve/main/starling-lm-7b-alpha.Q4_K_M.gguf", + "id": "starling-7b", + "object": "model", + "name": "Strarling alpha 7B", + "version": "1.0", + "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, + "system_prompt": "", + "user_prompt": "GPT4 User: ", + "ai_prompt": "<|end_of_turn|>\nGPT4 Assistant: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Berkeley-nest, The Bloke", + "tags": ["General", "Code"], + "size": 4370000000 + } + } + \ No newline at end of file diff --git a/models/storytelling-70b/model.json b/models/storytelling-70b/model.json new file mode 100644 index 000000000..76e6f7922 --- /dev/null +++ b/models/storytelling-70b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/GOAT-70B-Storytelling-GGUF/resolve/main/goat-70b-storytelling.Q5_K_M.gguf", + "id": "storytelling-70b", + "object": "model", + "name": "Storytelling 70B", + "version": "1.0", + "description": "The GOAT-70B-Storytelling model is designed for autonomous story-writing, including crafting books and movie scripts. Based on the LLaMA 2 70B architecture, this model excels in generating cohesive and engaging narratives using inputs like plot outlines and character profiles.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "### USER: ", + "ai_prompt": "\n### ASSISTANT: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "GOAT-AI, The Bloke", + "tags": ["General Use", "Writing"], + "size": 48750000000 + } + } + \ No newline at end of file diff --git a/models/tiefighter-13b/model.json b/models/tiefighter-13b/model.json new file mode 100644 index 000000000..b1d354ce3 --- /dev/null +++ b/models/tiefighter-13b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/LLaMA2-13B-Tiefighter-GGUF/resolve/main/llama2-13b-tiefighter.Q5_K_M.gguf", + "id": "tiefighter-13b", + "object": "model", + "name": "Tiefighter 13B", + "version": "1.0", + "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, + "system_prompt": "", + "user_prompt": "### Instruction: ", + "ai_prompt": "\n### Response: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "KoboldAI, The Bloke", + "tags": ["General Use", "Role-playing", "Writing"], + "size": 9230000000 + } + } + \ No newline at end of file diff --git a/models/tinyllama-1.1b/model.json b/models/tinyllama-1.1b/model.json new file mode 100644 index 000000000..f561eb25d --- /dev/null +++ b/models/tinyllama-1.1b/model.json @@ -0,0 +1,23 @@ +{ + "source_url": "https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v0.6/resolve/main/ggml-model-q4_0.gguf", + "id": "tinyllama-1.1b", + "object": "model", + "name": "TinyLlama Chat 1.1B", + "version": "1.0", + "description": "The TinyLlama project, featuring a 1.1B parameter Llama model, is pretrained on an expansive 3 trillion token dataset. Its design ensures easy integration with various Llama-based open-source projects. Despite its smaller size, it efficiently utilizes lower computational and memory resources, drawing on GPT-4's analytical prowess to enhance its conversational abilities and versatility.", + "format": "gguf", + "settings": { + "ctx_len": 2048, + "system_prompt": "<|system|>\n", + "user_prompt": "<|user|>\n", + "ai_prompt": "<|assistant|>\n" + }, + "parameters": { + "max_tokens": 2048 + }, + "metadata": { + "author": "TinyLlama", + "tags": ["General Use"], + "size": 637000000 + } +} \ No newline at end of file diff --git a/models/wizardcoder-13b/model.json b/models/wizardcoder-13b/model.json new file mode 100644 index 000000000..944b5632b --- /dev/null +++ b/models/wizardcoder-13b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/WizardCoder-Python-13B-V1.0-GGUF/resolve/main/wizardcoder-python-13b-v1.0.Q5_K_M.gguf", + "id": "wizardcoder-13b", + "object": "model", + "name": "Wizard Coder Python 13B", + "version": "1.0", + "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, + "system_prompt": "", + "user_prompt": "### Instruction:\n", + "ai_prompt": "### Response:\n" + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "WizardLM, The Bloke", + "tags": ["Code", "Big Context Length"], + "size": 9230000000 + } + } + \ No newline at end of file diff --git a/models/wizardcoder-34b/model.json b/models/wizardcoder-34b/model.json new file mode 100644 index 000000000..aa2618e1b --- /dev/null +++ b/models/wizardcoder-34b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/WizardCoder-Python-34B-V1.0-GGUF/resolve/main/wizardcoder-python-34b-v1.0.Q5_K_M.gguf", + "id": "wizardcoder-34b", + "object": "model", + "name": "Wizard Coder Python 34B", + "version": "1.0", + "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, + "system_prompt": "", + "user_prompt": "### Instruction:\n", + "ai_prompt": "### Response:\n" + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "WizardLM, The Bloke", + "tags": ["Code", "Big Context Length"], + "size": 24320000000 + } + } + \ No newline at end of file diff --git a/models/xwin-70b/model.json b/models/xwin-70b/model.json new file mode 100644 index 000000000..a5c1647b0 --- /dev/null +++ b/models/xwin-70b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Xwin-LM-70B-V0.1-GGUF/resolve/main/xwin-lm-70b-v0.1.Q5_K_M.gguf", + "id": "xwin-70b", + "object": "model", + "name": "Xwin LM 70B", + "version": "1.0", + "description": "Xwin-LM, based on Llama2 models, emphasizes alignment and exhibits advanced language understanding, text generation, and role-playing abilities.", + "format": "gguf", + "settings": { + "ctx_len": 4096, + "system_prompt": "", + "user_prompt": "USER: ", + "ai_prompt": "ASSISTANT: " + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "Xwin-LM, The Bloke", + "tags": ["General Use", "Role-playing"], + "size": 48750000000 + } + } + \ No newline at end of file diff --git a/models/yarn-70b/model.json b/models/yarn-70b/model.json new file mode 100644 index 000000000..67d8d3804 --- /dev/null +++ b/models/yarn-70b/model.json @@ -0,0 +1,21 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Yarn-Llama-2-70B-32k-GGUF/resolve/main/yarn-llama-2-70b-32k.Q5_K_M.gguf", + "id": "yarn-70b", + "object": "model", + "name": "Yarn 32k 70B", + "version": "1,0", + "description": "Yarn-Llama-2-70b-32k is designed specifically for handling long contexts. It represents an extension of the Llama-2-70b-hf model, now supporting a 32k token context window.", + "format": "gguf", + "settings": { + "ctx_len": 4096 + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "NousResearch, The Bloke", + "tags": ["General Use", "Big Context Length"], + "size": 48750000000 + } + } + \ No newline at end of file diff --git a/models/yi-34b/model.json b/models/yi-34b/model.json new file mode 100644 index 000000000..f899bc54b --- /dev/null +++ b/models/yi-34b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/Yi-34B-Chat-GGUF/resolve/main/yi-34b-chat.Q5_K_M.gguf", + "id": "yi-34b", + "object": "model", + "name": "Yi 34B", + "version": "1.0", + "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, + "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 + }, + "metadata": { + "author": "01-ai, The Bloke", + "tags": ["General", "Role-playing", "Writing"], + "size": 24320000000 + } + } + \ No newline at end of file diff --git a/models/zephyr-beta-7b/model.json b/models/zephyr-beta-7b/model.json new file mode 100644 index 000000000..24529bc9a --- /dev/null +++ b/models/zephyr-beta-7b/model.json @@ -0,0 +1,24 @@ +{ + "source_url": "https://huggingface.co/TheBloke/zephyr-7B-beta-GGUF/resolve/main/zephyr-7b-beta.Q4_K_M.gguf", + "id": "zephyr-beta-7b", + "object": "model", + "name": "Zephyr Beta 7B", + "version": "1.0", + "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, + "system_prompt": "<|system|>\n", + "user_prompt": "\n<|user|>\n", + "ai_prompt": "\n<|assistant|>\n" + }, + "parameters": { + "max_tokens": 4096 + }, + "metadata": { + "author": "HuggingFaceH4, The Bloke", + "tags": ["General Use", "Big Context Length"], + "size": 4370000000 + } + } + \ No newline at end of file From 1bf4c1b6210bdcc98ebf9ac0275fc870795041b7 Mon Sep 17 00:00:00 2001 From: NamH Date: Fri, 1 Dec 2023 18:03:51 +0700 Subject: [PATCH 2/2] feat: pre-populate Jan's /models folder (#796) Signed-off-by: James Co-authored-by: James --- .gitignore | 1 + core/src/core.ts | 4 + core/src/extensions/model.ts | 4 +- core/src/fs.ts | 4 + core/src/types/index.ts | 79 +++------ electron/handlers/app.ts | 8 +- electron/handlers/download.ts | 11 +- electron/handlers/extension.ts | 25 ++- electron/handlers/fs.ts | 11 +- electron/handlers/update.ts | 2 +- electron/invokers/fs.ts | 14 ++ electron/main.ts | 6 +- electron/managers/extension.ts | 4 +- electron/managers/module.ts | 2 +- electron/package.json | 7 +- electron/utils/path.ts | 15 ++ extensions/inference-extension/src/index.ts | 1 - .../model-extension/src/@types/global.d.ts | 5 +- .../model-extension/src/@types/schema.ts | 21 --- .../src/helpers/modelParser.ts | 46 ------ extensions/model-extension/src/index.ts | 151 +++++++++++++----- extensions/model-extension/webpack.config.js | 3 - package.json | 11 +- web/containers/ItemCardSidebar/index.tsx | 11 +- .../BottomBar/DownloadingState/index.tsx | 8 +- .../Layout/TopBar/CommandSearch/index.tsx | 25 +-- web/containers/ModalCancelDownload/index.tsx | 26 ++- web/containers/Providers/EventListener.tsx | 14 +- web/hooks/useActiveModel.ts | 3 +- web/hooks/useCreateNewThread.ts | 20 +++ web/hooks/useDeleteConversation.ts | 4 +- web/hooks/useDownloadModel.ts | 3 +- web/hooks/useDownloadState.ts | 24 +-- web/hooks/useGetConfiguredModels.ts | 20 +-- web/hooks/useGetMostSuitableModelVersion.ts | 27 ---- web/hooks/useGetPerformanceTag.ts | 50 ------ web/hooks/useSendChatMessage.ts | 47 +----- web/hooks/useSetActiveThread.ts | 5 - web/screens/Chat/MessageToolbar/index.tsx | 6 +- web/screens/Chat/Sidebar/index.tsx | 18 ++- .../ExploreModels/ExploreModelItem/index.tsx | 73 ++------- .../ExploreModelItemHeader/index.tsx | 54 ++----- .../ExploreModels/ExploreModelList/index.tsx | 8 +- .../ExploreModels/ModelVersionItem/index.tsx | 15 +- web/screens/MyModels/BlankState/index.tsx | 2 +- web/screens/MyModels/index.tsx | 5 +- web/screens/Settings/index.tsx | 1 - web/types/downloadState.d.ts | 1 - web/utils/dummy.ts | 119 -------------- 49 files changed, 362 insertions(+), 662 deletions(-) delete mode 100644 extensions/model-extension/src/@types/schema.ts delete mode 100644 extensions/model-extension/src/helpers/modelParser.ts delete mode 100644 web/hooks/useGetMostSuitableModelVersion.ts delete mode 100644 web/hooks/useGetPerformanceTag.ts delete mode 100644 web/utils/dummy.ts diff --git a/.gitignore b/.gitignore index a273404c2..4bfb0576f 100644 --- a/.gitignore +++ b/.gitignore @@ -10,6 +10,7 @@ dist build .DS_Store electron/renderer +electron/models package-lock.json *.log diff --git a/core/src/core.ts b/core/src/core.ts index b593277cb..0e032f4d9 100644 --- a/core/src/core.ts +++ b/core/src/core.ts @@ -54,6 +54,9 @@ const getUserSpace = (): Promise => window.core.api?.getUserSpace(); const openFileExplorer: (path: string) => Promise = (path) => window.core.api?.openFileExplorer(path); +const getResourcePath: () => Promise = () => + window.core.api?.getResourcePath(); + /** * Register extension point function type definition */ @@ -74,4 +77,5 @@ export { appDataPath, getUserSpace, openFileExplorer, + getResourcePath, }; diff --git a/core/src/extensions/model.ts b/core/src/extensions/model.ts index 3a5cc1ba3..276d15dcc 100644 --- a/core/src/extensions/model.ts +++ b/core/src/extensions/model.ts @@ -1,5 +1,5 @@ import { BaseExtension } from "../extension"; -import { Model, ModelCatalog } from "../types/index"; +import { Model } from "../types/index"; /** * Model extension for managing models. @@ -43,5 +43,5 @@ export abstract class ModelExtension extends BaseExtension { * Gets a list of configured models. * @returns A Promise that resolves with an array of configured models. */ - abstract getConfiguredModels(): Promise; + abstract getConfiguredModels(): Promise; } diff --git a/core/src/fs.ts b/core/src/fs.ts index e8eb38e04..d12b473bf 100644 --- a/core/src/fs.ts +++ b/core/src/fs.ts @@ -62,6 +62,9 @@ const deleteFile: (path: string) => Promise = (path) => const appendFile: (path: string, data: string) => Promise = (path, data) => window.core.api?.appendFile(path, data); +const copyFile: (src: string, dest: string) => Promise = (src, dest) => + window.core.api?.copyFile(src, dest); + /** * Reads a file line by line. * @param {string} path - The path of the file to read. @@ -80,4 +83,5 @@ export const fs = { deleteFile, appendFile, readLineByLine, + copyFile, }; diff --git a/core/src/types/index.ts b/core/src/types/index.ts index 15e83772f..bbd1e98de 100644 --- a/core/src/types/index.ts +++ b/core/src/types/index.ts @@ -180,7 +180,7 @@ export interface Model { /** * The version of the model. */ - version: string; + version: number; /** * The model download source. It can be an external url or a local filepath. @@ -197,12 +197,6 @@ export interface Model { */ name: string; - /** - * The organization that owns the model (you!) - * Default: "you" - */ - owned_by: string; - /** * The Unix timestamp (in seconds) for when the model was created */ @@ -236,11 +230,16 @@ export interface Model { metadata: ModelMetadata; } +export type ModelMetadata = { + author: string; + tags: string[]; + size: number; +}; + /** * The Model transition states. */ export enum ModelState { - ToDownload = "to_download", Downloading = "downloading", Ready = "ready", Running = "running", @@ -250,65 +249,27 @@ export enum ModelState { * The available model settings. */ export type ModelSettingParams = { - ctx_len: number; - ngl: number; - embedding: boolean; - n_parallel: number; + ctx_len?: number; + ngl?: number; + embedding?: boolean; + n_parallel?: number; + system_prompt?: string; + user_prompt?: string; + ai_prompt?: string; }; /** * The available model runtime parameters. */ export type ModelRuntimeParam = { - temperature: number; - token_limit: number; - top_k: number; - top_p: number; - stream: boolean; + temperature?: number; + token_limit?: number; + top_k?: number; + top_p?: number; + stream?: boolean; + max_tokens?: number; }; -/** - * The metadata of the model. - */ -export type ModelMetadata = { - engine: string; - quantization: string; - size: number; - binaries: string[]; - maxRamRequired: number; - author: string; - avatarUrl: string; -}; - -/** - * Model type of the presentation object which will be presented to the user - * @data_transfer_object - */ -export interface ModelCatalog { - /** The unique id of the model.*/ - id: string; - /** The name of the model.*/ - name: string; - /** The avatar url of the model.*/ - avatarUrl: string; - /** The short description of the model.*/ - shortDescription: string; - /** The long description of the model.*/ - longDescription: string; - /** The author name of the model.*/ - author: string; - /** The version of the model.*/ - version: string; - /** The origin url of the model repo.*/ - modelUrl: string; - /** The timestamp indicating when this model was released.*/ - releaseDate: number; - /** The tags attached to the model description **/ - tags: string[]; - /** The available versions of this model to download. */ - availableVersions: Model[]; -} - /** * Assistant type defines the shape of an assistant object. * @stored diff --git a/electron/handlers/app.ts b/electron/handlers/app.ts index 4a6d56b6a..adbc875b2 100644 --- a/electron/handlers/app.ts +++ b/electron/handlers/app.ts @@ -1,9 +1,9 @@ import { app, ipcMain, shell } from 'electron' -import { ModuleManager } from '../managers/module' +import { ModuleManager } from './../managers/module' import { join } from 'path' -import { ExtensionManager } from '../managers/extension' -import { WindowManager } from '../managers/window' -import { userSpacePath } from '../utils/path' +import { ExtensionManager } from './../managers/extension' +import { WindowManager } from './../managers/window' +import { userSpacePath } from './../utils/path' export function handleAppIPCs() { /** diff --git a/electron/handlers/download.ts b/electron/handlers/download.ts index 316576e89..1776fccd9 100644 --- a/electron/handlers/download.ts +++ b/electron/handlers/download.ts @@ -1,9 +1,10 @@ import { app, ipcMain } from 'electron' -import { DownloadManager } from '../managers/download' +import { DownloadManager } from './../managers/download' import { resolve, join } from 'path' -import { WindowManager } from '../managers/window' +import { WindowManager } from './../managers/window' import request from 'request' -import { createWriteStream, unlink } from 'fs' +import { createWriteStream } from 'fs' +import { getResourcePath } from './../utils/path' const progress = require('request-progress') export function handleDownloaderIPCs() { @@ -37,6 +38,10 @@ export function handleDownloaderIPCs() { rq?.abort() }) + ipcMain.handle('getResourcePath', async (_event) => { + return getResourcePath() + }) + /** * Downloads a file from a given URL. * @param _event - The IPC event object. diff --git a/electron/handlers/extension.ts b/electron/handlers/extension.ts index 1af1be36c..5c2c13ff4 100644 --- a/electron/handlers/extension.ts +++ b/electron/handlers/extension.ts @@ -1,19 +1,16 @@ -import { app, ipcMain, webContents } from 'electron' -import { readdirSync, rmdir, writeFileSync } from 'fs' -import { ModuleManager } from '../managers/module' +import { ipcMain, webContents } from 'electron' +import { readdirSync } from 'fs' +import { ModuleManager } from './../managers/module' import { join, extname } from 'path' -import { ExtensionManager } from '../managers/extension' -import { WindowManager } from '../managers/window' -import { manifest, tarball } from 'pacote' import { getActiveExtensions, getAllExtensions, installExtensions, -} from '../extension/store' -import { getExtension } from '../extension/store' -import { removeExtension } from '../extension/store' -import Extension from '../extension/extension' -import { userSpacePath } from '../utils/path' +} from './../extension/store' +import { getExtension } from './../extension/store' +import { removeExtension } from './../extension/store' +import Extension from './../extension/extension' +import { getResourcePath, userSpacePath } from './../utils/path' export function handleExtensionIPCs() { /**MARK: General handlers */ @@ -48,11 +45,7 @@ export function handleExtensionIPCs() { * @returns An array of paths to the base extensions. */ ipcMain.handle('extension:baseExtensions', async (_event) => { - const baseExtensionPath = join( - __dirname, - '../', - app.isPackaged ? '../../app.asar.unpacked/pre-install' : '../pre-install' - ) + const baseExtensionPath = join(getResourcePath(), 'pre-install') return readdirSync(baseExtensionPath) .filter((file) => extname(file) === '.tgz') .map((file) => join(baseExtensionPath, file)) diff --git a/electron/handlers/fs.ts b/electron/handlers/fs.ts index 9c39c1092..16cef6eb6 100644 --- a/electron/handlers/fs.ts +++ b/electron/handlers/fs.ts @@ -1,8 +1,9 @@ -import { app, ipcMain } from 'electron' +import { ipcMain } from 'electron' import * as fs from 'fs' +import fse from 'fs-extra' import { join } from 'path' import readline from 'readline' -import { userSpacePath } from '../utils/path' +import { userSpacePath } from './../utils/path' /** * Handles file system operations. @@ -145,6 +146,12 @@ export function handleFsIPCs() { } }) + ipcMain.handle('copyFile', async (_event, src: string, dest: string) => { + console.debug(`Copying file from ${src} to ${dest}`) + + return fse.copySync(src, dest, { overwrite: false }) + }) + /** * Reads a file line by line. * @param event - The event object. diff --git a/electron/handlers/update.ts b/electron/handlers/update.ts index 340db54b9..08d32fffe 100644 --- a/electron/handlers/update.ts +++ b/electron/handlers/update.ts @@ -1,5 +1,5 @@ import { app, dialog } from "electron"; -import { WindowManager } from "../managers/window"; +import { WindowManager } from "./../managers/window"; import { autoUpdater } from "electron-updater"; export function handleAppUpdates() { diff --git a/electron/invokers/fs.ts b/electron/invokers/fs.ts index d7d204d0a..309562ad6 100644 --- a/electron/invokers/fs.ts +++ b/electron/invokers/fs.ts @@ -67,6 +67,20 @@ export function fsInvokers() { * @param {string} path - The path of the directory to remove. */ rmdir: (path: string) => ipcRenderer.invoke('rmdir', path), + + /** + * Copies a file from the source path to the destination path. + * @param {string} src - The source path of the file to copy. + * @param {string} dest - The destination path where the file should be copied. + */ + copyFile: (src: string, dest: string) => ipcRenderer.invoke('copyFile', src, dest), + + /** + * Retrieves the resource path. + * @returns {Promise} A promise that resolves to the resource path. + */ + getResourcePath: () => ipcRenderer.invoke('getResourcePath'), + } return interfaces diff --git a/electron/main.ts b/electron/main.ts index cfd6ca665..189836866 100644 --- a/electron/main.ts +++ b/electron/main.ts @@ -1,7 +1,7 @@ import { app, BrowserWindow } from 'electron' import { join } from 'path' import { setupMenu } from './utils/menu' -import { handleFsIPCs } from './handlers/fs' +import { createUserSpace, getResourcePath } from './utils/path' /** * Managers @@ -18,9 +18,11 @@ import { handleThemesIPCs } from './handlers/theme' import { handleExtensionIPCs } from './handlers/extension' import { handleAppIPCs } from './handlers/app' import { handleAppUpdates } from './handlers/update' +import { handleFsIPCs } from './handlers/fs' app .whenReady() + .then(createUserSpace) .then(ExtensionManager.instance.migrateExtensions) .then(ExtensionManager.instance.setupExtensions) .then(setupMenu) @@ -56,7 +58,7 @@ function createMainWindow() { }) const startURL = app.isPackaged - ? `file://${join(__dirname, '../renderer/index.html')}` + ? `file://${join(__dirname, '..', 'renderer', 'index.html')}` : 'http://localhost:3000' /* Load frontend app to the window */ diff --git a/electron/managers/extension.ts b/electron/managers/extension.ts index e23c75ddf..7eef24877 100644 --- a/electron/managers/extension.ts +++ b/electron/managers/extension.ts @@ -1,10 +1,10 @@ import { app } from 'electron' -import { init } from '../extension' +import { init } from './../extension' import { join, resolve } from 'path' import { rmdir } from 'fs' import Store from 'electron-store' import { existsSync, mkdirSync, writeFileSync } from 'fs' -import { userSpacePath } from '../utils/path' +import { userSpacePath } from './../utils/path' /** * Manages extension installation and migration. */ diff --git a/electron/managers/module.ts b/electron/managers/module.ts index 43dda0fb6..dc16d0d22 100644 --- a/electron/managers/module.ts +++ b/electron/managers/module.ts @@ -1,4 +1,4 @@ -import { dispose } from "../utils/disposable"; +import { dispose } from "./../utils/disposable"; /** * Manages imported modules. diff --git a/electron/package.json b/electron/package.json index 46e9b328c..627f5ad54 100644 --- a/electron/package.json +++ b/electron/package.json @@ -13,10 +13,12 @@ "renderer/**/*", "build/*.{js,map}", "build/**/*.{js,map}", - "pre-install" + "pre-install", + "models/**/*" ], "asarUnpack": [ - "pre-install" + "pre-install", + "models" ], "publish": [ { @@ -70,6 +72,7 @@ "@uiball/loaders": "^1.3.0", "electron-store": "^8.1.0", "electron-updater": "^6.1.4", + "fs-extra": "^11.2.0", "pacote": "^17.0.4", "request": "^2.88.2", "request-progress": "^3.0.0", diff --git a/electron/utils/path.ts b/electron/utils/path.ts index 30eb0ef2d..8f3092561 100644 --- a/electron/utils/path.ts +++ b/electron/utils/path.ts @@ -1,4 +1,19 @@ import { join } from 'path' import { app } from 'electron' +import { mkdir } from 'fs-extra' + +export async function createUserSpace(): Promise { + return mkdir(userSpacePath).catch(() => {}) +} export const userSpacePath = join(app.getPath('home'), 'jan') + +export function getResourcePath() { + let appPath = join(app.getAppPath(), '..', 'app.asar.unpacked') + + if (!app.isPackaged) { + // for development mode + appPath = join(__dirname, '..', '..') + } + return appPath +} diff --git a/extensions/inference-extension/src/index.ts b/extensions/inference-extension/src/index.ts index 1ba471ab1..e8e7758bb 100644 --- a/extensions/inference-extension/src/index.ts +++ b/extensions/inference-extension/src/index.ts @@ -146,7 +146,6 @@ export default class JanInferenceExtension implements InferenceExtension { object: "thread.message", }; events.emit(EventName.OnMessageResponse, message); - console.log(JSON.stringify(data, null, 2)); instance.isCancelled = false; instance.controller = new AbortController(); diff --git a/extensions/model-extension/src/@types/global.d.ts b/extensions/model-extension/src/@types/global.d.ts index 87056c342..bb030c762 100644 --- a/extensions/model-extension/src/@types/global.d.ts +++ b/extensions/model-extension/src/@types/global.d.ts @@ -1,3 +1,2 @@ -declare const PLUGIN_NAME: string; -declare const MODULE_PATH: string; -declare const MODEL_CATALOG_URL: string; +declare const PLUGIN_NAME: string +declare const MODULE_PATH: string diff --git a/extensions/model-extension/src/@types/schema.ts b/extensions/model-extension/src/@types/schema.ts deleted file mode 100644 index 1d3c3a7d1..000000000 --- a/extensions/model-extension/src/@types/schema.ts +++ /dev/null @@ -1,21 +0,0 @@ -interface Version { - name: string - quantMethod: string - bits: number - size: number - maxRamRequired: number - usecase: string - downloadLink: string -} -interface ModelSchema { - id: string - name: string - shortDescription: string - avatarUrl: string - longDescription: string - author: string - version: string - modelUrl: string - tags: string[] - versions: Version[] -} diff --git a/extensions/model-extension/src/helpers/modelParser.ts b/extensions/model-extension/src/helpers/modelParser.ts deleted file mode 100644 index 3a397fb7d..000000000 --- a/extensions/model-extension/src/helpers/modelParser.ts +++ /dev/null @@ -1,46 +0,0 @@ -import { ModelCatalog } from '@janhq/core' - -export const parseToModel = (modelGroup): ModelCatalog => { - const modelVersions = [] - modelGroup.versions.forEach((v) => { - const model = { - object: 'model', - version: modelGroup.version, - source_url: v.downloadLink, - id: v.name, - name: v.name, - owned_by: 'you', - created: 0, - description: modelGroup.longDescription, - state: 'to_download', - settings: v.settings, - parameters: v.parameters, - metadata: { - engine: '', - quantization: v.quantMethod, - size: v.size, - binaries: [], - maxRamRequired: v.maxRamRequired, - author: modelGroup.author, - avatarUrl: modelGroup.avatarUrl, - }, - } - modelVersions.push(model) - }) - - const modelCatalog: ModelCatalog = { - id: modelGroup.id, - name: modelGroup.name, - avatarUrl: modelGroup.avatarUrl, - shortDescription: modelGroup.shortDescription, - longDescription: modelGroup.longDescription, - author: modelGroup.author, - version: modelGroup.version, - modelUrl: modelGroup.modelUrl, - releaseDate: modelGroup.createdAt, - tags: modelGroup.tags, - availableVersions: modelVersions, - } - - return modelCatalog -} diff --git a/extensions/model-extension/src/index.ts b/extensions/model-extension/src/index.ts index a2b0be304..d0267b84e 100644 --- a/extensions/model-extension/src/index.ts +++ b/extensions/model-extension/src/index.ts @@ -1,6 +1,12 @@ -import { ExtensionType, fs, downloadFile, abortDownload } from '@janhq/core' -import { ModelExtension, Model, ModelCatalog } from '@janhq/core' -import { parseToModel } from './helpers/modelParser' +import { + ExtensionType, + fs, + downloadFile, + abortDownload, + getResourcePath, + getUserSpace, +} from '@janhq/core' +import { ModelExtension, Model, ModelState } from '@janhq/core' import { join } from 'path' /** @@ -24,10 +30,7 @@ export default class JanModelExtension implements ModelExtension { * @override */ onLoad(): void { - /** Cloud Native - * TODO: Fetch all downloading progresses? - **/ - fs.mkdir(JanModelExtension._homeDir) + this.copyModelsToHomeDir() } /** @@ -36,6 +39,30 @@ export default class JanModelExtension implements ModelExtension { */ onUnload(): void {} + private async copyModelsToHomeDir() { + try { + // list all of the files under the home directory + const files = await fs.listFiles('') + + if (files.includes(JanModelExtension._homeDir)) { + // ignore if the model is already downloaded + console.debug('Model already downloaded') + return + } + + // copy models folder from resources to home directory + const resourePath = await getResourcePath() + const srcPath = join(resourePath, 'models') + + const userSpace = await getUserSpace() + const destPath = join(userSpace, JanModelExtension._homeDir) + + await fs.copyFile(srcPath, destPath) + } catch (err) { + console.error(err) + } + } + /** * Downloads a machine learning model. * @param model - The model to download. @@ -57,11 +84,11 @@ export default class JanModelExtension implements ModelExtension { * @returns {Promise} A promise that resolves when the download has been cancelled. */ async cancelModelDownload(modelId: string): Promise { - return abortDownload(join(JanModelExtension._homeDir, modelId, modelId)).then( - () => { - fs.rmdir(join(JanModelExtension._homeDir, modelId)) - } - ) + return abortDownload( + join(JanModelExtension._homeDir, modelId, modelId) + ).then(() => { + fs.deleteFile(join(JanModelExtension._homeDir, modelId, modelId)) + }) } /** @@ -72,7 +99,26 @@ export default class JanModelExtension implements ModelExtension { async deleteModel(modelId: string): Promise { try { const dirPath = join(JanModelExtension._homeDir, modelId) - await fs.rmdir(dirPath) + + // remove all files under dirPath except model.json + const files = await fs.listFiles(dirPath) + const deletePromises = files.map((fileName: string) => { + if (fileName !== JanModelExtension._modelMetadataFileName) { + return fs.deleteFile(join(dirPath, fileName)) + } + }) + await Promise.allSettled(deletePromises) + + // update the state as default + const jsonFilePath = join( + dirPath, + JanModelExtension._modelMetadataFileName + ) + const json = await fs.readFile(jsonFilePath) + const model = JSON.parse(json) as Model + delete model.state + + await fs.writeFile(jsonFilePath, JSON.stringify(model, null, 2)) } catch (err) { console.error(err) } @@ -91,7 +137,17 @@ export default class JanModelExtension implements ModelExtension { ) try { - await fs.writeFile(jsonFilePath, JSON.stringify(model, null, 2)) + await fs.writeFile( + jsonFilePath, + JSON.stringify( + { + ...model, + state: ModelState.Ready, + }, + null, + 2 + ) + ) } catch (err) { console.error(err) } @@ -102,39 +158,62 @@ export default class JanModelExtension implements ModelExtension { * @returns A Promise that resolves with an array of all models. */ async getDownloadedModels(): Promise { - const results: Model[] = [] - const allDirs: string[] = await fs.listFiles(JanModelExtension._homeDir) - for (const dir of allDirs) { - const modelDirPath = join(JanModelExtension._homeDir, dir) - const isModelDir = await fs.isDirectory(modelDirPath) - if (!isModelDir) { - // if not a directory, ignore - continue + const models = await this.getModelsMetadata() + return models.filter((model) => model.state === ModelState.Ready) + } + + private async getModelsMetadata(): Promise { + try { + const filesUnderJanRoot = await fs.listFiles('') + if (!filesUnderJanRoot.includes(JanModelExtension._homeDir)) { + console.debug('model folder not found') + return [] } - const jsonFiles: string[] = (await fs.listFiles(modelDirPath)).filter( - (fileName: string) => fileName === JanModelExtension._modelMetadataFileName - ) + const files: string[] = await fs.listFiles(JanModelExtension._homeDir) - for (const json of jsonFiles) { - const model: Model = JSON.parse( - await fs.readFile(join(modelDirPath, json)) + const allDirectories: string[] = [] + for (const file of files) { + const isDirectory = await fs.isDirectory( + join(JanModelExtension._homeDir, file) ) - results.push(model) + if (isDirectory) { + allDirectories.push(file) + } } - } - return results + const readJsonPromises = allDirectories.map((dirName) => { + const jsonPath = join( + JanModelExtension._homeDir, + dirName, + JanModelExtension._modelMetadataFileName + ) + return this.readModelMetadata(jsonPath) + }) + const results = await Promise.allSettled(readJsonPromises) + const modelData = results.map((result) => { + if (result.status === 'fulfilled') { + return JSON.parse(result.value) as Model + } else { + console.error(result.reason) + } + }) + return modelData + } catch (err) { + console.error(err) + return [] + } + } + + private readModelMetadata(path: string) { + return fs.readFile(join(path)) } /** * Gets all available models. * @returns A Promise that resolves with an array of all models. */ - getConfiguredModels(): Promise { - // Add a timestamp to the URL to prevent caching - return import( - /* webpackIgnore: true */ MODEL_CATALOG_URL + `?t=${Date.now()}` - ).then((module) => module.default.map((e) => parseToModel(e))) + async getConfiguredModels(): Promise { + return this.getModelsMetadata() } } diff --git a/extensions/model-extension/webpack.config.js b/extensions/model-extension/webpack.config.js index 3475516ed..a9332da99 100644 --- a/extensions/model-extension/webpack.config.js +++ b/extensions/model-extension/webpack.config.js @@ -19,9 +19,6 @@ module.exports = { new webpack.DefinePlugin({ PLUGIN_NAME: JSON.stringify(packageJson.name), MODULE_PATH: JSON.stringify(`${packageJson.name}/${packageJson.module}`), - MODEL_CATALOG_URL: JSON.stringify( - 'https://cdn.jsdelivr.net/npm/@janhq/models@latest/dist/index.js' - ), }), ], output: { diff --git a/package.json b/package.json index a2476887f..9192a0238 100644 --- a/package.json +++ b/package.json @@ -6,8 +6,7 @@ "uikit", "core", "electron", - "web", - "server" + "web" ], "nohoist": [ "uikit", @@ -17,15 +16,13 @@ "electron", "electron/**", "web", - "web/**", - "server", - "server/**" + "web/**" ] }, "scripts": { "lint": "yarn workspace jan lint && yarn workspace jan-web lint", "test": "yarn workspace jan test:e2e", - "dev:electron": "yarn workspace jan dev", + "dev:electron": "cpx \"models/**\" \"electron/models/\" && yarn workspace jan dev", "dev:web": "yarn workspace jan-web dev", "dev": "concurrently --kill-others \"yarn dev:web\" \"wait-on http://localhost:3000 && yarn dev:electron\"", "test-local": "yarn lint && yarn build:test && yarn test", @@ -33,7 +30,7 @@ "build:uikit": "yarn workspace @janhq/uikit install && yarn workspace @janhq/uikit build", "build:core": "cd core && yarn install && yarn run build", "build:web": "yarn workspace jan-web build && cpx \"web/out/**\" \"electron/renderer/\"", - "build:electron": "yarn workspace jan build", + "build:electron": "yarn workspace jan build && cpx \"models/**\" \"electron/models/\"", "build:electron:test": "yarn workspace jan build:test", "build:extensions": "rimraf ./electron/pre-install/*.tgz && concurrently --kill-others-on-fail \"cd ./extensions/conversational-extension && npm install && npm run build:publish\" \"cd ./extensions/inference-extension && npm install && npm run build:publish\" \"cd ./extensions/model-extension && npm install && npm run build:publish\" \"cd ./extensions/monitoring-extension && npm install && npm run build:publish\" \"cd ./extensions/assistant-extension && npm install && npm run build:publish\"", "build:test": "yarn build:web && yarn workspace jan build:test", diff --git a/web/containers/ItemCardSidebar/index.tsx b/web/containers/ItemCardSidebar/index.tsx index b6a7bacbd..627d7f45d 100644 --- a/web/containers/ItemCardSidebar/index.tsx +++ b/web/containers/ItemCardSidebar/index.tsx @@ -1,9 +1,16 @@ type Props = { title: string description?: string + disabled?: boolean + onChange?: (text?: string) => void } -export default function ItemCardSidebar({ description, title }: Props) { +export default function ItemCardSidebar({ + description, + title, + disabled, + onChange, +}: Props) { return (
@@ -11,9 +18,11 @@ export default function ItemCardSidebar({ description, title }: Props) {
onChange?.(e.target.value)} />
) diff --git a/web/containers/Layout/BottomBar/DownloadingState/index.tsx b/web/containers/Layout/BottomBar/DownloadingState/index.tsx index bc456bf80..1aad0fb1c 100644 --- a/web/containers/Layout/BottomBar/DownloadingState/index.tsx +++ b/web/containers/Layout/BottomBar/DownloadingState/index.tsx @@ -69,18 +69,14 @@ export default function DownloadingState() { />
-

{item?.fileName}

+

{item?.modelId}

{formatDownloadPercentage(item?.percent)}
+ {activeThread && ( + + )} ) } diff --git a/web/containers/ModalCancelDownload/index.tsx b/web/containers/ModalCancelDownload/index.tsx index de54e1cf8..8619c543c 100644 --- a/web/containers/ModalCancelDownload/index.tsx +++ b/web/containers/ModalCancelDownload/index.tsx @@ -24,34 +24,30 @@ import { extensionManager } from '@/extension' import { downloadingModelsAtom } from '@/helpers/atoms/Model.atom' type Props = { - suitableModel: Model + model: Model isFromList?: boolean } -export default function ModalCancelDownload({ - suitableModel, - isFromList, -}: Props) { +export default function ModalCancelDownload({ model, isFromList }: Props) { const { modelDownloadStateAtom } = useDownloadState() const downloadAtom = useMemo( - () => atom((get) => get(modelDownloadStateAtom)[suitableModel.name]), + () => atom((get) => get(modelDownloadStateAtom)[model.id]), // eslint-disable-next-line react-hooks/exhaustive-deps - [suitableModel.name] + [model.id] ) const models = useAtomValue(downloadingModelsAtom) const downloadState = useAtomValue(downloadAtom) + const cancelText = `Cancel ${formatDownloadPercentage(downloadState.percent)}` return ( {isFromList ? ( ) : ( - + )} @@ -60,7 +56,7 @@ export default function ModalCancelDownload({

Are you sure you want to cancel the download of  - {downloadState?.fileName}? + {downloadState?.modelId}?

@@ -71,11 +67,7 @@ export default function ModalCancelDownload({ - - {show && ( - - )} -
- )}
) diff --git a/web/screens/ExploreModels/ExploreModelItemHeader/index.tsx b/web/screens/ExploreModels/ExploreModelItemHeader/index.tsx index c845c5a45..b51ec164c 100644 --- a/web/screens/ExploreModels/ExploreModelItemHeader/index.tsx +++ b/web/screens/ExploreModels/ExploreModelItemHeader/index.tsx @@ -1,7 +1,7 @@ /* eslint-disable react-hooks/exhaustive-deps */ -import { useCallback, useEffect, useMemo, useState } from 'react' +import { useCallback, useMemo, useState } from 'react' -import { Model, ModelCatalog } from '@janhq/core' +import { Model } from '@janhq/core' import { Badge, Button } from '@janhq/uikit' import { atom, useAtomValue } from 'jotai' @@ -15,67 +15,41 @@ import { ModelPerformance, TagType } from '@/constants/tagType' import useDownloadModel from '@/hooks/useDownloadModel' import { useDownloadState } from '@/hooks/useDownloadState' import { useGetDownloadedModels } from '@/hooks/useGetDownloadedModels' -import useGetPerformanceTag from '@/hooks/useGetPerformanceTag' import { useMainViewState } from '@/hooks/useMainViewState' import { toGigabytes } from '@/utils/converter' -import { totalRamAtom } from '@/helpers/atoms/SystemBar.atom' - type Props = { - suitableModel: Model - exploreModel: ModelCatalog + model: Model } -const ExploreModelItemHeader: React.FC = ({ - suitableModel, - exploreModel, -}) => { +const ExploreModelItemHeader: React.FC = ({ model }) => { const { downloadModel } = useDownloadModel() const { downloadedModels } = useGetDownloadedModels() const { modelDownloadStateAtom, downloadStates } = useDownloadState() - const { getPerformanceForModel } = useGetPerformanceTag() const [title, setTitle] = useState('Recommended') - const totalRam = useAtomValue(totalRamAtom) + const [performanceTag, setPerformanceTag] = useState( ModelPerformance.PerformancePositive ) const downloadAtom = useMemo( - () => atom((get) => get(modelDownloadStateAtom)[suitableModel.name]), - [suitableModel.name] + () => atom((get) => get(modelDownloadStateAtom)[model.id]), + [model.id] ) const downloadState = useAtomValue(downloadAtom) const { setMainViewState } = useMainViewState() - const calculatePerformance = useCallback( - (suitableModel: Model) => async () => { - const { title, performanceTag } = await getPerformanceForModel( - suitableModel, - totalRam - ) - setPerformanceTag(performanceTag) - setTitle(title) - }, - [totalRam] - ) - - useEffect(() => { - calculatePerformance(suitableModel) - }, [suitableModel]) - const onDownloadClick = useCallback(() => { - downloadModel(suitableModel) + downloadModel(model) // eslint-disable-next-line react-hooks/exhaustive-deps - }, [suitableModel]) + }, [model]) - // TODO: Comparing between Model Id and Version Name? - const isDownloaded = - downloadedModels.find((model) => model.id === suitableModel.name) != null + const isDownloaded = downloadedModels.find((md) => md.id === model.id) != null let downloadButton = ( ) @@ -93,7 +67,7 @@ const ExploreModelItemHeader: React.FC = ({ } if (downloadState != null && downloadStates.length > 0) { - downloadButton = + downloadButton = } const renderBadge = (performance: TagType) => { @@ -115,7 +89,7 @@ const ExploreModelItemHeader: React.FC = ({ return (
- {exploreModel.name} + {model.name} {performanceTag && renderBadge(performanceTag)}
{downloadButton} diff --git a/web/screens/ExploreModels/ExploreModelList/index.tsx b/web/screens/ExploreModels/ExploreModelList/index.tsx index 8c0c9bdb4..eea9f0238 100644 --- a/web/screens/ExploreModels/ExploreModelList/index.tsx +++ b/web/screens/ExploreModels/ExploreModelList/index.tsx @@ -1,16 +1,14 @@ -import { ModelCatalog } from '@janhq/core' +import { Model } from '@janhq/core' import ExploreModelItem from '@/screens/ExploreModels/ExploreModelItem' type Props = { - models: ModelCatalog[] + models: Model[] } const ExploreModelList: React.FC = ({ models }) => (
- {models?.map((item, i) => ( - - ))} + {models?.map((model) => )}
) diff --git a/web/screens/ExploreModels/ModelVersionItem/index.tsx b/web/screens/ExploreModels/ModelVersionItem/index.tsx index f7d09307b..e16c477f6 100644 --- a/web/screens/ExploreModels/ModelVersionItem/index.tsx +++ b/web/screens/ExploreModels/ModelVersionItem/index.tsx @@ -2,7 +2,7 @@ import React, { useMemo } from 'react' import { Model } from '@janhq/core' -import { Badge, Button } from '@janhq/uikit' +import { Button } from '@janhq/uikit' import { atom, useAtomValue } from 'jotai' import ModalCancelDownload from '@/containers/ModalCancelDownload' @@ -63,7 +63,7 @@ const ModelVersionItem: React.FC = ({ model }) => { } if (downloadState != null && downloadStates.length > 0) { - downloadButton = + downloadButton = } return ( @@ -74,16 +74,7 @@ const ModelVersionItem: React.FC = ({ model }) => {
-
- {`${toGigabytes( - model.metadata.maxRamRequired - )} RAM required`} - {toGigabytes(model.metadata.size)} -
+
{downloadButton}
diff --git a/web/screens/MyModels/BlankState/index.tsx b/web/screens/MyModels/BlankState/index.tsx index a820440d0..c0d7be6bb 100644 --- a/web/screens/MyModels/BlankState/index.tsx +++ b/web/screens/MyModels/BlankState/index.tsx @@ -55,7 +55,7 @@ export default function BlankStateMyModel() { } />
-

{item?.fileName}

+

{item?.modelId}

{formatDownloadPercentage(item?.percent)}
diff --git a/web/screens/MyModels/index.tsx b/web/screens/MyModels/index.tsx index d9c2d2880..c8176f010 100644 --- a/web/screens/MyModels/index.tsx +++ b/web/screens/MyModels/index.tsx @@ -63,10 +63,7 @@ const MyModelsScreen = () => {
- + {model.metadata.author.charAt(0)} diff --git a/web/screens/Settings/index.tsx b/web/screens/Settings/index.tsx index ced7589b5..63c343add 100644 --- a/web/screens/Settings/index.tsx +++ b/web/screens/Settings/index.tsx @@ -30,7 +30,6 @@ const SettingsScreen = () => { setMenus(menu) }, []) - const preferenceExtensions = preferenceItems .map((x) => x.extensionnName) .filter((x, i) => { diff --git a/web/types/downloadState.d.ts b/web/types/downloadState.d.ts index cb154522d..3c3389b4f 100644 --- a/web/types/downloadState.d.ts +++ b/web/types/downloadState.d.ts @@ -4,7 +4,6 @@ type DownloadState = { speed: number percent: number size: DownloadSize - fileName: string error?: string } diff --git a/web/utils/dummy.ts b/web/utils/dummy.ts deleted file mode 100644 index bde61e38f..000000000 --- a/web/utils/dummy.ts +++ /dev/null @@ -1,119 +0,0 @@ -/* eslint-disable @typescript-eslint/naming-convention */ -import { ModelCatalog, ModelState } from '@janhq/core' - -export const dummyModel: ModelCatalog = { - id: 'aladar/TinyLLama-v0-GGUF', - name: 'TinyLLama-v0-GGUF', - shortDescription: 'TinyLlama-1.1B-Chat-v0.3-GGUF', - longDescription: 'https://huggingface.co/aladar/TinyLLama-v0-GGUF/tree/main', - avatarUrl: '', - releaseDate: Date.now(), - author: 'aladar', - version: '1.0.0', - modelUrl: 'aladar/TinyLLama-v0-GGUF', - tags: ['freeform', 'tags'], - availableVersions: [ - { - object: 'model', - version: '1.0.0', - source_url: - 'https://huggingface.co/aladar/TinyLLama-v0-GGUF/resolve/main/TinyLLama-v0.Q8_0.gguf', - id: 'TinyLLama-v0.Q8_0.gguf', - name: 'TinyLLama-v0.Q8_0.gguf', - owned_by: 'you', - created: 0, - description: '', - state: ModelState.ToDownload, - settings: { - ctx_len: 2048, - ngl: 100, - embedding: true, - n_parallel: 4, - }, - parameters: { - temperature: 0.7, - token_limit: 2048, - top_k: 0, - top_p: 1, - stream: true, - }, - metadata: { - engine: '', - quantization: '', - size: 5816320, - binaries: [], - maxRamRequired: 256000000, - author: 'aladar', - avatarUrl: '', - }, - }, - { - object: 'model', - version: '1.0.0', - source_url: - 'https://huggingface.co/aladar/TinyLLama-v0-GGUF/resolve/main/TinyLLama-v0.f16.gguf', - id: 'TinyLLama-v0.f16.gguf', - name: 'TinyLLama-v0.f16.gguf', - owned_by: 'you', - created: 0, - description: '', - state: ModelState.ToDownload, - settings: { - ctx_len: 2048, - ngl: 100, - embedding: true, - n_parallel: 4, - }, - parameters: { - temperature: 0.7, - token_limit: 2048, - top_k: 0, - top_p: 1, - stream: true, - }, - metadata: { - engine: '', - quantization: '', - size: 5816320, - binaries: [], - maxRamRequired: 256000000, - author: 'aladar', - avatarUrl: '', - }, - }, - { - object: 'model', - version: '1.0.0', - source_url: - 'https://huggingface.co/aladar/TinyLLama-v0-GGUF/resolve/main/TinyLLama-v0.f32.gguf', - id: 'TinyLLama-v0.f32.gguf', - name: 'TinyLLama-v0.f32.gguf', - owned_by: 'you', - created: 0, - description: '', - state: ModelState.ToDownload, - settings: { - ctx_len: 2048, - ngl: 100, - embedding: true, - n_parallel: 4, - }, - parameters: { - temperature: 0.7, - token_limit: 2048, - top_k: 0, - top_p: 1, - stream: true, - }, - metadata: { - engine: '', - quantization: '', - size: 5816320, - binaries: [], - maxRamRequired: 256000000, - author: 'aladar', - avatarUrl: '', - }, - }, - ], -}