fix(models): update ctxlen + maxlength to 4096

models/capybara-34b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "USER:\n{prompt}\nASSISTANT:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "NousResearch, The Bloke",

models/deepseek-coder-1.3b/model.json

@@ -8,11 +8,11 @@
     "description": "Deepseek Coder trained on 2T tokens (87% code, 13% English/Chinese), excelling in project-level code completion with advanced capabilities across multiple programming languages.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### Instruction:\n{prompt}\n### Response:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Deepseek, The Bloke",

models/deepseek-coder-34b/model.json

@@ -7,11 +7,11 @@
     "description": "Deepseek Coder trained on 2T tokens (87% code, 13% English/Chinese), excelling in project-level code completion with advanced capabilities across multiple programming languages.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### Instruction:\n{prompt}\n### Response:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Deepseek, The Bloke",

models/llama2-chat-70b-q4/model.json

@@ -7,11 +7,11 @@
     "description": "This is a 4-bit quantized version of Meta AI's Llama 2 Chat 70b model.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "[INST] <<SYS>>\n{system_message}<</SYS>>\n{prompt}[/INST]"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "MetaAI, The Bloke",

models/llama2-chat-7b-q4/model.json

@@ -7,11 +7,11 @@
     "description": "This is a 4-bit quantized iteration of Meta AI's Llama 2 Chat 7b model, specifically designed for a comprehensive understanding through training on extensive internet data.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "[INST] <<SYS>>\n{system_message}<</SYS>>\n{prompt}[/INST]"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "MetaAI, The Bloke",

models/lzlv-70b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "USER:\n{prompt}\nASSISTANT:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Lizpreciatior, The Bloke",

models/mistral-ins-7b-q4/model.json

@@ -7,14 +7,14 @@
     "description": "This is a 4-bit quantized iteration of MistralAI's Mistral Instruct 7b model, specifically designed for a comprehensive understanding through training on extensive internet data.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "system_prompt": "",
       "user_prompt": "<s>[INST]",
       "ai_prompt": "[/INST]",
       "prompt_template": "<s>[INST]{prompt}\n[/INST]"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "MistralAI, The Bloke",

models/mixtral-8x7b-instruct/model.json

@@ -7,11 +7,11 @@
     "description": "The Mixtral-8x7B Large Language Model (LLM) is a pretrained generative Sparse Mixture of Experts. The Mixtral-8x7B outperforms Llama 2 70B on most benchmarks we tested.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "[INST] {prompt} [/INST]"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "MistralAI, TheBloke",

models/noromaid-20b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### Instruction:{prompt}\n### Response:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "NeverSleep, The Bloke",

models/openhermes-neural-7b/model.json

@@ -7,11 +7,11 @@
     "description": "OpenHermes Neural is a merged model using the TIES method.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Intel, Jan",

models/pandora-10.7b-v1/model.json

@@ -7,11 +7,11 @@
     "description": "Pandora, our research model, employs the Passthrough merging technique to merge 2x7B models into 1.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Jan",

models/phind-34b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### System Prompt\n{system_message}\n### User Message\n{prompt}\n### Assistant"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Phind, The Bloke",

models/solar-10.7b-instruct/model.json

@@ -7,11 +7,11 @@
     "description": "SOLAR-10.7B model built on the Llama2 architecture with Depth Up-Scaling and integrated Mistral 7B weights. Its robustness and adaptability make it ideal for fine-tuning applications, significantly enhancing performance with simple instruction-based techniques.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### User: {prompt}\n### Assistant:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Upstage, Jan",

models/solar-10.7b-slerp/model.json

@@ -7,11 +7,11 @@
     "description": "This model uses the Slerp merge method from SOLAR Instruct and Pandora-v1",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### User: {prompt}\n### Assistant:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Jan",

models/starling-7b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "GPT4 User: {prompt}<|end_of_turn|>GPT4 Assistant:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Berkeley-nest, The Bloke",

models/tinyllama-1.1b/model.json

@@ -7,11 +7,11 @@
   "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,
+      "ctx_len": 4096,
       "prompt_template": "<|system|>\n{system_message}<|user|>\n{prompt}<|assistant|>"
   },
   "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
   },
   "metadata": {
       "author": "TinyLlama",

models/trinity-v1-7b/model.json

@@ -7,11 +7,11 @@
     "description": "Trinity is an experimental model merge of GreenNodeLM & LeoScorpius using the Slerp method. Recommended for daily assistance purposes.",
     "format": "gguf",
     "settings": {
-      "ctx_len": 2048,
+      "ctx_len": 4096,
       "prompt_template": "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "Jan",

models/wizardcoder-13b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "### Instruction:\n{prompt}\n### Response:"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "WizardLM, The Bloke",

models/yi-34b/model.json

@@ -7,11 +7,11 @@
     "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": 2048,
+      "ctx_len": 4096,
       "prompt_template": "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
     },
     "parameters": {
-      "max_tokens": 2048
+      "max_tokens": 4096
     },
     "metadata": {
       "author": "01-ai, The Bloke",