下载SmolLM2模型并进行推理的简单示例

最新推荐文章于 2025-03-14 15:52:33 发布

C7211BA

最新推荐文章于 2025-03-14 15:52:33 发布

阅读量513

点赞数 15

文章标签：大语言模型 kaggle huggingface llm

版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。

本文链接：https://blog.csdn.net/C7211BA/article/details/143483605

版权

代码说明

在kaggle平台上下载huggingface平台上的HuggingFaceTB/SmolLM2-1.7B-Instruct模型，并使用kaggle平台提供的免费gpu资源进行推理的简单示例。
在这里插入图片描述

源代码

from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "HuggingFaceTB/SmolLM2-1.7B-Instruct"

device = "cuda" # for GPU usage or "cpu" for CPU usage
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
# for multiple GPUs install accelerate and do `model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map="auto")`
model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device)

messages = [{"role": "user", "content": "What is the capital of France."}]
input_text=tokenizer.apply_chat_template(messages, tokenize=False)
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
outputs = model.generate(inputs, max_new_tokens=50, temperature=0.2, top_p=0.9, do_sample=True)
print(tokenizer.decode(outputs[0]))

在这里插入图片描述

博客等级

码龄7年

102
原创

955
点赞

501
收藏

521
粉丝

关注

私信

热门文章

最新评论

基于python的随机迷宫游戏
2501_91331098: import pygame import random import sys # 初始化参数 WIDTH, HEIGHT = 400, 500 GRID_SIZE = 8 CELL_SIZE = 50 COLORS = [(255,0,0), (0,255,0), (0,0,255), (255,255,0), (128,0,128)] class Gem: def __init__(self, x, y): self.rect = pygame.Rect(x, y, CELL_SIZE-2, CELL_SIZE-2) self.color = random.choice(COLORS) self.target_y = y self.sparkle = random.randint(0, 10) # 闪烁效果参数 class Match3Game: def __init__(self): pygame.init() self.screen = pygame.display.set_mode((WIDTH, HEIGHT)) pygame.display.set_caption("PyMatch3") self.clock = pygame.time.Clock() self.font = pygame.font.SysFont(None, 36) # 游戏状态初始化 self.grid = [[None for _ in range(GRID_SIZE)] for _ in range(GRID_SIZE)] self.selected = None self.score = 0 self.combo = 0 self.level = 1 self.required_score = 5000 # 初始化网格 for y in range(GRID_SIZE): for x in range(GRID
基于python的随机迷宫游戏
2501_90899552: import pygame import sys import random # 初始化Pygame pygame.init() # 定义颜色 WHITE = (255, 255, 255) BLACK = (0, 0, 0) RED = (255, 0, 0) GREEN = (0, 255, 0) BLUE = (0, 0, 255) # 设置屏幕大小 WIDTH, HEIGHT = 400, 400 screen = pygame.display.set_mode((WIDTH, HEIGHT)) pygame.display.set_caption("随机迷宫游戏") # 定义方块大小 MAZE_SIZE = 8 # 迷宫大小 (8x8) BLOCK_SIZE = WIDTH // MAZE_SIZE # 定义玩家初始位置 player_pos = [1, 1] # 生成随机迷宫 def generate_maze(size): # 初始化迷宫，全部为墙壁 maze = [[1 for _ in range(size)] for _ in range(size)] # 使用深度优先搜索算法生成迷宫 def dfs(x, y): maze[y][x] = 0 # 打通当前单元格 directions = [(1, 0), (-1, 0), (0, 1), (0, -1)] random.shuffle(directions) # 随机打乱方向 for dx, dy in directions: nx, ny = x + dx * 2, y + dy * 2 if 0 <= nx < size and 0 <= ny < size and maze[ny][nx] == 1: maze[y + dy][x + dx] = 0 # 打通路径 dfs(nx, ny) dfs(1, 1) # 从起点开始生成迷宫 maze[1][1] = 0 # 确保起点是空的 maze[size - 2][size - 2] = 2 # 设置出口 return maze # 重置游戏状态 def reset_game(): global maze, player_pos maze = generate_maze(MAZE_SIZE) player_pos = [1, 1] # 游戏主循环 def main(): clock = pygame.time.Clock() running = True reset_game() # 初始化迷宫和玩家位置 while running: for event in pygame.event.get(): if event.type == pygame.QUIT: running = False elif event.type == pygame.KEYDOWN: # 处理玩家移动 if event.key == pygame.K_UP: move_player(0, -1) elif event.key == pygame.K_DOWN: move_player(0, 1) elif event.key == pygame.K_LEFT: move_player(-1, 0) elif event.key == pygame.K_RIGHT: move_player(1, 0) # 绘制迷宫和玩家 screen.fill(WHITE) draw_maze() draw_player() pygame.display.flip() # 检查是否到达出口 if maze[player_pos[1]][player_pos[0]] == 2: print("恭喜你，成功找到出口！进入下一局...") reset_game() # 重置游戏状态 clock.tick(30) pygame.quit() sys.exit() # 绘制迷宫 def draw_maze(): for y, row in enumerate(maze): for x, block in enumerate(row): if block == 1: pygame.draw.rect(screen, BLACK, (x * BLOCK_SIZE, y * BLOCK_SIZE, BLOCK_SIZE, BLOCK_SIZE)) elif block == 2: pygame.draw.rect(screen, GREEN, (x * BLOCK_SIZE, y * BLOCK_SIZE, BLOCK_SIZE, BLOCK_SIZE)) # 绘制玩家 def draw_player(): pygame.draw.circle(screen, RED, (player_pos[0] * BLOCK_SIZE + BLOCK_SIZE // 2, player_pos[1] * BLOCK_SIZE + BLOCK_SIZE // 2), BLOCK_SIZE // 3) # 移动玩家 def move_player(dx, dy): new_x = player_pos[0] + dx new_y = player_pos[1] + dy # 检查新位置是否合法 if 0 <= new_x < len(maze[0]) and 0 <= new_y < len(maze): if maze[new_y][new_x] != 1: # 不能穿过墙壁 player_pos[0] = new_x player_pos[1] = new_y if __name__ == "__main__": main()
降维算法概述
小ᶻ☡꙳ᵃⁱᵍᶜ꙳: 内容很赞，感谢您的分享
Accurate predictions on small data with a tabular foundation model
小ᶻ☡꙳ᵃⁱᵍᶜ꙳: 很喜欢这篇文章，向您学习了
图算法概述
小ᶻ☡꙳ᵃⁱᵍᶜ꙳: 内容很赞

大家在看

最新文章

目录

评论

被折叠的条评论为什么被折叠?

到【灌水乐园】发言

查看更多评论

添加红包

成就一亿技术人!

hope_wisdom

发出的红包

实付元

使用余额支付

点击重新获取

扫码支付

钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。