Investment Rating - The report maintains an "Outperform" rating for the media and internet sector [5][42]. Core Views - The report expresses optimism regarding opportunities in the gaming and film industries, particularly with the upcoming National Day holiday and the recent issuance of game licenses [1][4]. - The gaming sector is expected to benefit from a new product cycle, while the film industry is anticipated to experience a bottom reversal [4][42]. - AI applications are highlighted as a significant area for investment, with various companies making advancements in AI technologies [2][4][42]. Summary by Sections Industry Performance - The media sector saw a weekly increase of 0.47%, underperforming compared to the CSI 300 (1.15%) and the ChiNext Index (1.80%) [1][12]. - Notable gainers included Guomai Culture and Guiguang Network, while Jin Yi Film and China Film experienced significant declines [1][12]. Key Developments - In September, over 100 game licenses were issued, including 145 domestic and 11 imported games [2][17]. - Major AI advancements were reported, including Alibaba's Wan2.5 model and Meta's Code World Model [2][19]. Box Office and Content Performance - The box office for the week of September 22-28 reached 354 million yuan, with "731" leading at 1.065 billion yuan, accounting for 87.9% of the total [3][20]. - Popular TV shows included "Earth Super Fresh" and "Flowers and Youth" [27][25]. Investment Recommendations - The report recommends focusing on gaming stocks like Kaiying Network and Jiubite, as well as media companies such as Mango Super Media and Bilibili [4][42]. - The report emphasizes the importance of AI applications across various sectors, including gaming, education, and advertising [4][42]. Company Earnings Forecasts - Key companies such as Kaiying Network and Mango Super Media are projected to have earnings per share (EPS) of 1.01 yuan and 0.81 yuan for 2025, respectively [5][45]. - The report provides a detailed valuation table for several companies, all rated as "Outperform" [5][45].

Core Insights - Meta's FAIR team has launched the Code World Model (CWM), a large language model (LLM) with 32 billion parameters and a context length of up to 131k tokens, aimed at integrating "world model" concepts into code generation and reasoning [1][2][3] - CWM is designed to not only write code but also simulate code execution, reason about program states, and self-detect and fix bugs, enhancing the model's understanding of code execution [2][3] Training Phases - The training of CWM is divided into three main phases: - Pre-training with 8 trillion tokens, where approximately 30% are code-related [3][4] - Mid-training, which incorporates 5 trillion tokens of world modeling data, extending the context length to 131k tokens [4][6] - Post-training (SFT + RL), involving 100 billion tokens for instruction and reasoning capabilities, followed by large-scale multi-task reinforcement learning with 172 billion tokens [4][10] Data Utilization - CWM's world model capabilities are driven by two main types of data during mid-training: - Execution traces from Python, which help the model learn how code execution alters local states [6][8] - Interaction trajectories from an automated agent that executes tasks in a repository, collecting around 3 million trajectories from 10.2k images and 3.15k repositories [9] Performance Metrics - In benchmark tests, CWM demonstrated strong performance, achieving 65.8% pass@1 on SWE-bench Verified with Test-Time-Scaling enabled, and notable results on LiveCodeBench (68.6%), Math-500 (96.6%), and AIME 2024 (76.0%) [10][12] - CWM's performance is competitive with larger or closed-source LLMs, nearing GPT-4 levels, although it has limitations in certain editing formats and multi-language scenarios [12] Industry Reception - The release of CWM has garnered significant attention, with Meta's AI researchers actively promoting it, highlighting its potential impact on software development [13][15] - While the open-sourcing of CWM's training checkpoints is praised for its utility in academic and engineering replication, there are concerns regarding the model's computational demands and the need for practical testing in real development environments [15]