repo_name stringlengths 1 62 | dataset stringclasses 1
value | lang stringclasses 11
values | pr_id int64 1 20.1k | owner stringlengths 2 34 | reviewer stringlengths 2 39 | diff_hunk stringlengths 15 262k | code_review_comment stringlengths 1 99.6k |
|---|---|---|---|---|---|---|---|
neural-speed | github_2023 | cpp | 11 | intel | zhewang1-intc | @@ -398,19 +406,24 @@ class gemm_t<
matB, matB_payload);
subgroup::tile_load<cache_hint::cached, cache_hint::cached>(
scale, scale_payload);
- subgroup::tile_load<cache_hint::cached, cache_hint::cached>(
- zero_pt, zero_pt_payload);
+ ... | in fact, jblas suppor asym+s4_fullrange config(but qbits only expose sym+bitN feature to user), so it's ok to load zp when quant_mode==fullrange(if user choose asym quant alg) in xetla level |
neural-speed | github_2023 | cpp | 11 | intel | luoyu-intel | @@ -389,7 +399,9 @@ class gemm_t<
if constexpr (enable_periodic_sync) {
if ((i % sync_freq) == 0) {
if constexpr (wg_size_x > 1) { nbarrier_a.arrive(); }
- if constexpr (wg_size_y > 1) { nbarrier_b.arrive(); }
+ if constexpr (arch_... | any benefit here? It seems only two barriers' sync points, and no other interactive codes after sync. |
neural-speed | github_2023 | cpp | 11 | intel | luoyu-intel | @@ -771,8 +299,13 @@ class gemm_universal_t<dispatch_policy_int4_dequantize_kslicing<group_swizzle_,
__XETLA_API static constexpr uint32_t get_slm_size() {
constexpr uint32_t size = gemm_slm_size * num_local_kslicing
+ kslicing_slm_size + epilogue_slm_size * num_local_kslicing;
- s... | as a 'xxx_xe.hpp', should arch_tag be fixed one in this file like L128 does? |
neural-speed | github_2023 | cpp | 11 | intel | luoyu-intel | @@ -839,45 +372,54 @@ class gemm_universal_t<dispatch_policy_int4_dequantize_kslicing<group_swizzle_,
bool implementable = true;
if (gemm_t::msg_type_a != msg_type::unaligned_2d) {
if (gemm_t::msg_type_a == msg_type::block_2d) {
- implementable &= kernel::block_2d<gpu_arch:... | L128 change arch_tag to gpu_arch:Xe, why opposite here? |
neural-speed | github_2023 | cpp | 11 | intel | zhewang1-intc | @@ -478,512 +499,48 @@ class gemm_t<
auto scale_vec
= scale.reg.xetla_select<scale_t::block_size_x, 1>(
scale_block_id * scale_t::block_size_x);
- auto zero_pt_vec
- = zero_pt.reg
- ... | seems that we unpack a int4-value and move it 4bit to a uint8_t's low-4-bit postion.
so it's not the `S4_CLIP` definition in Jblas(we keep the high-4-bit position in Jblas).
I guess this is what S4_ASYM definition in inc |
neural-speed | github_2023 | others | 25 | intel | airMeng | @@ -110,7 +110,7 @@ if(UT_BUILD)
add_executable(${PROJECT_NAME}_ut ${srcs} ${headers} ${ut_headers})
if(BTLA_UT_OPENMP)
include(FindOpenMP)
- target_link_libraries(${PROJECT_NAME}_ut PRIVATE OpenMP::OpenMP_CXX OpenMP::OpenMP_C)
+ target_link_libraries(${PROJECT_NAME}_ut PRIVATE OpenMP::OpenMP_CXX) | OpenMP_C is for GGML operators, need to link OpenMP_C for GGML somewhere else |
neural-speed | github_2023 | cpp | 12 | intel | zhewang1-intc | @@ -888,10 +903,10 @@ inline void f32_f4_quantize_4x16(const float* srcptr, int8_t* dstptr, int ld_src
zmm2 = _mm512_add_ps(zmm2, zmm_zp);
zmm3 = _mm512_add_ps(zmm3, zmm_zp);
} else {
- mask4 = _mm512_cmplt_ps_mask(zmm0, zmm_v0);
- mask5 = _mm512_cmplt_ps_mask(zmm1, zmm_v0);
- mask6 = _mm512_cmplt... | why there are 4 different imm8? |
neural-speed | github_2023 | others | 12 | intel | zhewang1-intc | @@ -59,11 +59,7 @@ option(NE_AVX512_VBMI "neural_engine: enable AVX512-VBMI"
option(NE_AVX512_VNNI "neural_engine: enable AVX512-VNNI" OFF)
option(NE_FMA "neural_engine: enable FMA" ON)
option(NE_AMX ... | quite confused about neural_engine here, it should already been deprecated in deep-speed. |
neural-speed | github_2023 | others | 12 | intel | zhewang1-intc | @@ -59,11 +59,7 @@ option(NE_AVX512_VBMI "neural_engine: enable AVX512-VBMI"
option(NE_AVX512_VNNI "neural_engine: enable AVX512-VNNI" OFF)
option(NE_FMA "neural_engine: enable FMA" ON)
option(NE_AMX ... | consider to clean the cmakelist? we no longer dependent on oneDNN |
neural-speed | github_2023 | cpp | 18 | intel | DDEle | @@ -1843,7 +1843,7 @@ void jblas_fusion_attn_int8_forward(const attn_int8_fwd_args_t* params) {
}
size_t jblas_fusion_attn_workspace_size(const attn_shape_t* params) {
const auto& p = *params; // TODO(Yi): Better way to get tmp size?
- return size_t(omp_get_max_threads() * sizeof(float) * 16) * padto(padto(p.sl_... | Thanks for the correction. |
neural-speed | github_2023 | others | 14 | intel | airMeng | @@ -0,0 +1,50 @@
+# Copyright (C) 2023 Intel Corporation
+# SPDX-License-Identifier: BSD-3-Clause
+
+ARG IMAGE_NAME=ubuntu
+ARG IMAGE_TAG=22.04
+FROM ${IMAGE_NAME}:${IMAGE_TAG} as base
+
+FROM base as neural-speed
+
+ARG PYTHON=python3.10
+
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && apt-get install -y... | should be
```suggestion
cd /neural_speed && \
``` |
gprofiler-performance-studio | github_2023 | go | 5 | intel | leonid-granulate | @@ -262,11 +262,6 @@ func NewClickHouseClient(addr string) *ClickHouseClient {
log.Fatal(err)
}
if err := db.Ping(); err != nil {
- if exception, ok := err.(*clickhouse.Exception); ok { | Instead of remove just change them to log.Debug |
gprofiler-performance-studio | github_2023 | others | 4 | intel | leonid-granulate | @@ -32,7 +32,7 @@ require (
github.com/modern-go/reflect2 v1.0.2 // indirect
github.com/montanaflynn/stats v0.6.6 // indirect
github.com/ugorji/go/codec v1.2.11 // indirect
- golang.org/x/crypto v0.9.0 // indirect
+ golang.org/x/crypto v0.17.0 // indirect | did you tried to rebuild after a version change? |
gprofiler-performance-studio | github_2023 | javascript | 4 | intel | shirmaor | @@ -15,7 +17,7 @@ import Flexbox from '@/components/common/layout/Flexbox';
import { FgContext, SelectorsContext } from '@/states';
import { FilterTagsContext } from '@/states/filters/FiltersTagsContext';
import { COLORS } from '@/theme/colors';
-import { PROFILES_VIEWS } from '@/utils/consts';
+import {PROFILES_VIE... | ```suggestion
import { PROFILES_VIEWS } from '@/utils/consts';
``` |
gprofiler-performance-studio | github_2023 | javascript | 4 | intel | shirmaor | @@ -7,6 +7,8 @@
import { Box, ListItemButton, ListItemIcon, Menu } from '@mui/material';
import { useContext, useState } from 'react';
+import {useLocation} from "react-router-dom"; | ```suggestion
import { useLocation } from "react-router-dom";
``` |
gprofiler-performance-studio | github_2023 | javascript | 4 | intel | shirmaor | @@ -7,6 +7,8 @@
import { Box, ListItemButton, ListItemIcon, Menu } from '@mui/material';
import { useContext, useState } from 'react';
+import {useLocation} from "react-router-dom";
+ | ```suggestion
``` |
gprofiler-performance-studio | github_2023 | others | 3 | intel | timson | @@ -0,0 +1,262 @@
+-- INTEL CONFIDENTIAL
+-- Copyright (C) 2023 Intel Corporation
+-- This software and the related documents are Intel copyrighted materials, and your use of them is governed by the express license under which they were provided to you ("License"). Unless the License provides otherwise, you may not use... | ```suggestion
-- Sharding key is CallStackHash
``` |
gprofiler-performance-studio | github_2023 | others | 3 | intel | timson | @@ -0,0 +1,262 @@
+-- INTEL CONFIDENTIAL
+-- Copyright (C) 2023 Intel Corporation
+-- This software and the related documents are Intel copyrighted materials, and your use of them is governed by the express license under which they were provided to you ("License"). Unless the License provides otherwise, you may not use... | maybe remove this? |
tsffs | github_2023 | others | 98 | intel | riverar | @@ -77,15 +77,16 @@ tracing = { version = "0.1.40", features = ["log"] }
yaxpeax-arm = "0.3.0"
chrono = "0.4.38"
vergilius = "0.1.2"
-windows = { version = "0.57.0", features = [
- "Win32_Foundation",
+windows-sys = { features = [
+ "Win32",
"Win32_System",
"Win32_System_SystemServices",
"Win32_... | Can reduce some of the duplicity here, thanks to the crate's feature hierarchy. For example, `Win32_System_Diagnostics_Debug` brings in `Win32_System_Diagnostics` which brings in `Win32_System`..., etc.
```toml
[dependencies.windows-sys]
version = "0.59.0"
features = [
"Win32_Foundation",
"Win32_System_Dia... |
tsffs | github_2023 | others | 85 | intel | tklengyel | @@ -0,0 +1,108 @@
+#!/bin/bash
+
+# Copyright (C) 2024 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+# NOTE: This script requires the Intel version of `ispm` to be installed and available
+# on the PATH. We do not download a new copy of it because despite its public
+# unavailability, we want to be able t... | Should do hash checks on these downloaded files. |
tsffs | github_2023 | others | 38 | intel | antoniogi | @@ -1 +1,10 @@
-# Running the Fuzzer
\ No newline at end of file
+# Running the Fuzzer
+
+Once a fuzzing campaign is set up, you can generally run the fuzzer like:
+
+```sh
+./simics -no-gui --no-win --batch-mode fuzz.simics
+```
+
+At a log level (`tsffs.log-level 2`) of `2` or greater, you'll see statistics of the
+c... | To me, it reads better like this:
At a log level of `2` or greater (ie `tsffs.log-level 2`) , you'll see statistics of the current progress during execution. |
tsffs | github_2023 | others | 82 | intel | kalvdans | @@ -102,5 +102,5 @@ Brandon Marken Ph.D.
<brandon.marken@intel.com>
Robert Geunzel Ph.D. | ```suggestion
Robert Guenzel Ph.D.
``` |
tsffs | github_2023 | others | 75 | intel | tklengyel | @@ -60,68 +37,224 @@ where
/// SIMICS version of the low level bindings. This is not needed for all consumers of the
/// API, but is useful for consumers which need to remain compatible with a wide range of
/// SIMICS base versions.
-pub fn emit_cfg_directives() -> anyhow::Result<()> {
+///
+/// Changelog: | I think you want to put this as a separate Markdown file in the docs folder so that it's easier to find/read. |
tsffs | github_2023 | others | 75 | intel | tklengyel | @@ -754,7 +754,7 @@ impl CInterface {
.join("py-typemaps.c"),
)
.arg(interface_subdir.join(&header_name))
- .print_args()
+ // .print_args() | Don't leave in commented out code. Either make it conditional on some debug flag/verbosity, or remove it. |
ipex-llm-tutorial | github_2023 | others | 83 | intel | plusbang | @@ -0,0 +1,131 @@
+# 6.2 Run Whisper (medium) on Intel GPUs
+
+You can use IPEX-LLM to load Transformer-based automatic speech recognition (ASR) models for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) for ASR can be loaded and optimized automatically on Intel GPUs with low-bit quantizat... | Add `FP6` in 'INT4/NF4/INT5/FP8/INT8' |
ipex-llm-tutorial | github_2023 | others | 83 | intel | plusbang | @@ -0,0 +1,131 @@
+# 6.2 Run Whisper (medium) on Intel GPUs
+
+You can use IPEX-LLM to load Transformer-based automatic speech recognition (ASR) models for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) for ASR can be loaded and optimized automatically on Intel GPUs with low-bit quantizat... | Please adjust the spacing. |
ipex-llm-tutorial | github_2023 | others | 83 | intel | plusbang | @@ -0,0 +1,131 @@
+# 6.2 Run Whisper (medium) on Intel GPUs
+
+You can use IPEX-LLM to load Transformer-based automatic speech recognition (ASR) models for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) for ASR can be loaded and optimized automatically on Intel GPUs with low-bit quantizat... | Please add some note here about the source of `audio_en.wav` (such as, from dataset voxpopuli?), and "audio_zh.wav" also needs. |
ipex-llm-tutorial | github_2023 | others | 82 | intel | Oscilloscope98 | @@ -5,7 +5,7 @@ You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs
In this tutorial, you will learn how to run LLMs on Intel GPUs with IPEX-LLM optimizations, and based on that build a stream chatbot. A popular open-source LLM [baichuan-inc/Baichuan2-7B-Chat](https://www.modelscope.cn/mod... | for modelscope usage -> to use ModelScope Library ? |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,108 @@
+# 6.1 Run Baichuan 2 (7B) on Intel GPUs | 6.1 -> 6.2, other subtitles are similar |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,108 @@
+# 6.1 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any Hugging Face *transformers* model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on Hugging Face can be loaded and optimized automatically on Intel GPUs with low-bit quantization (sup... | INT4/NF4/INT5/INT8 -> INT4/NF4/INT5/FP8/INT8 |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,108 @@
+# 6.1 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any Hugging Face *transformers* model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on Hugging Face can be loaded and optimized automatically on Intel GPUs with low-bit quantization (sup... | Please use modelscope hub (https://www.modelscope.cn/models/baichuan-inc/Baichuan2-7B-Chat/summary) to load model. |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,109 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | Please also add some comments about usage of this parameter :) |
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | `conduct **implicit** quantization while loading.` is a quite outdated explanation, just removed it :) |
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | Let's remove the `cpu_embedding` and `model_hub='modelscope'` explaination just in note box :) |
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | I think we supported more https://github.com/intel-analytics/ipex-llm/blob/main/python/llm/src/ipex_llm/transformers/__init__.py#L19
Maybe list them all, or list the most important onces and add `, etc.` to the and |
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | Please make sure we have list all options here |
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | Please also add some notes as below:
|
ipex-llm-tutorial | github_2023 | others | 80 | intel | Oscilloscope98 | @@ -0,0 +1,110 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | Maybe add a note here to remind reader that they need to prepare the env for IPEX-LLM on GPU first |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,117 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | some model has different model id in huggingface and modelscope, should we add some comment here? |
ipex-llm-tutorial | github_2023 | others | 80 | intel | plusbang | @@ -0,0 +1,117 @@
+# 6.2 Run Baichuan 2 (7B) on Intel GPUs
+
+You can use IPEX-LLM to load any ModelScope model for acceleration on Intel GPUs. With IPEX-LLM, PyTorch models (in FP16/BF16/FP32) hosted on ModelScope can be loaded and optimized automatically on Intel GPUs with low-bit quantization (supported precisions i... | should we add `model_hub` description here instead of in note box? |
ipex-llm-tutorial | github_2023 | others | 81 | intel | plusbang | @@ -0,0 +1,216 @@
+{
+ "cells": [
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 3: Basic Application Development On Baichuan2\n",
+ "\n",
+ "This notebook introduces the essential usage of `ipex-llm`, and walks you through building a very basic chat ... | 'take a relatively small LLM, i.e. [Baichuan2-7b-chat]' -> 'take [Baichuan2-7b-chat]' |
ipex-llm-tutorial | github_2023 | others | 81 | intel | plusbang | @@ -0,0 +1,216 @@
+{
+ "cells": [
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 3: Basic Application Development On Baichuan2\n",
+ "\n",
+ "This notebook introduces the essential usage of `ipex-llm`, and walks you through building a very basic chat ... | 'NF4/INT5/INT8' -> 'FP8/INT8' |
ipex-llm-tutorial | github_2023 | others | 81 | intel | Oscilloscope98 | @@ -0,0 +1,204 @@
+{
+ "cells": [
+ {
+ "attachments": {},
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 3: Basic Application Development On Baichuan2\n",
+ "\n",
+ "This notebook introduces the essential usage of `ipex-llm`, and walks you through building a very basic chat ... | @plusbang Should we also add a notebox for how to install on Linux OS? |
ipex-llm-tutorial | github_2023 | others | 81 | intel | plusbang | @@ -28,6 +28,15 @@
"!pip install --pre --upgrade ipex-llm[all]"
]
},
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "> **Note**\n",
+ ">\n",
+ "> Please refer to https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Overview/install_cpu.html#quick-installation for more details ... | It maybe better to remind users that:
"On Linux OS, we recommend to use `pip install --pre --upgrade ipex-llm[all] --extra-index-url https://download.pytorch.org/whl/cpu` to install. Please refer to https://ipex-llm.readthedocs.io/en/latest/doc/LLM/Overview/install_cpu.html#quick-installation for more details." |
ipex-llm-tutorial | github_2023 | others | 71 | intel | plusbang | @@ -22,13 +22,10 @@ In Chapter 6, you will learn how to run LLMs, as well as implement stream chat f
+ [Llama2-7B](./6_1_GPU_Llama2-7B.md)
-## 6.0 Environment Setup
-Here are some best practices for setting up your environment. It is strongly recommended that you follow the corresponding steps below to configure... | Chapter 7 -> Chapter 6 |
ipex-llm-tutorial | github_2023 | others | 71 | intel | plusbang | @@ -23,20 +23,24 @@ In Chapter 6, you will learn how to run LLMs, as well as implement stream chat f
+ [Llama2-7B](./6_1_GPU_Llama2-7B.md)
-## 6.0 System Recommendation
+## 6.0 System Support
+### 1. Linux:
+**Hardware**:
+- Intel Arc™ A-Series Graphics
+- Intel Data Center GPU Flex Series
+- Intel Data Center GP... | The two links of [HF-Transformers-AutoModels] and [PyTorch-Models] looks strange here.
Maybe we could describe: "For Windows, BigDL-LLM is verified to accelerate models with both `AutoModel` APIs and `optimize_model` API on Intel iGPU and dGPU." |
ipex-llm-tutorial | github_2023 | others | 71 | intel | plusbang | @@ -12,15 +12,16 @@ In Chapter 7, you will go through how to fine-tune a large language model to a t
We are going to train with a popular open source model [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) as an example.
-## 7.0 System Recommendation
+## 7.0 System Support
-> ⚠️Hardware
- - Intel... | We could remove `### Linux:` here. |
ipex-llm-tutorial | github_2023 | others | 71 | intel | Oscilloscope98 | @@ -22,81 +22,27 @@ In Chapter 6, you will learn how to run LLMs, as well as implement stream chat f
+ [Llama2-7B](./6_1_GPU_Llama2-7B.md)
-## 6.0 Environment Setup
-Here are some best practices for setting up your environment. It is strongly recommended that you follow the corresponding steps below to configure... | Let's remove line 37? |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -40,10 +40,12 @@ For a smooth experience with the notebooks in Chatper 7, please ensure your hard
> **Note**
> Please note that only Linux OS has been supported for BigDL-LLM optimizations on Intel GPUs. | This note is quite out of date. We also support Windows now. |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -77,16 +77,13 @@ conda init
> **Note**
> Python 3.9 is recommended for running BigDL-LLM.
-Create a Python 3.9 environment with the name you choose, for example `llm-tutorial-gpu`:
+Create a Python 3.9 environment with the name you choose, for example `llm-tutorial-gpu` and install `bigdl-llm` from PyPI:
```b... | Maybe we could add comment here to emphasize ipex2.0 is installed here. |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -1,7 +1,15 @@
## Environment setup for Intel Arc GPU
-For Linux users, Ubuntu 22.04 and Linux kernel 5.19.0 is prefered. Ubuntu 22.04 and Linux kernel 5.19.0-41-generic is mostly used in our test environment. But default linux kernel of ubuntu 22.04.3 is 6.2.0-35-generic, so we recommonded you to downgrade kernel t... | highlight: **Prerequisites** |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -1,7 +1,15 @@
## Environment setup for Intel Arc GPU
-For Linux users, Ubuntu 22.04 and Linux kernel 5.19.0 is prefered. Ubuntu 22.04 and Linux kernel 5.19.0-41-generic is mostly used in our test environment. But default linux kernel of ubuntu 22.04.3 is 6.2.0-35-generic, so we recommonded you to downgrade kernel t... | recommonded -> recommend |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -24,11 +24,11 @@ In Chapter 6, you will learn how to run LLMs, as well as implement stream chat f
## 6.0 Environment Setup
-Here are some best practices for setting up your environment. It is strongly recommended that you follow the corresponding steps below to configure your environment properly.
+**The follow... | I feel we could also add link to llm document here. |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -95,10 +106,167 @@ sudo apt install -y intel-oneapi-common-vars=2023.2.0-49462 \
intel-oneapi-tbb=2021.10.0-49541 intel-oneapi-tbb-devel=2021.10.0-49541\
intel-oneapi-ccl=2021.10.0-49084 intel-oneapi-ccl-devel=2021.10.0-49084\
intel-oneapi-dnnl-devel=2023.2.0-49516 intel-oneapi-dnnl=2023.2.0-49516
-# ... | highlight: **Prerequisites** |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -95,10 +106,167 @@ sudo apt install -y intel-oneapi-common-vars=2023.2.0-49462 \
intel-oneapi-tbb=2021.10.0-49541 intel-oneapi-tbb-devel=2021.10.0-49541\
intel-oneapi-ccl=2021.10.0-49084 intel-oneapi-ccl-devel=2021.10.0-49084\
intel-oneapi-dnnl-devel=2023.2.0-49516 intel-oneapi-dnnl=2023.2.0-49516
-# ... | highlight: **Importance** |
ipex-llm-tutorial | github_2023 | others | 67 | intel | plusbang | @@ -95,10 +106,167 @@ sudo apt install -y intel-oneapi-common-vars=2023.2.0-49462 \
intel-oneapi-tbb=2021.10.0-49541 intel-oneapi-tbb-devel=2021.10.0-49541\
intel-oneapi-ccl=2021.10.0-49084 intel-oneapi-ccl-devel=2021.10.0-49084\
intel-oneapi-dnnl-devel=2023.2.0-49516 intel-oneapi-dnnl=2023.2.0-49516
-# ... | I feel it maybe better to add comment here to emphasize `[xpu]` installation option on Windows corresponds to ipex2.1. |
ipex-llm-tutorial | github_2023 | others | 56 | intel | hkvision | @@ -105,10 +105,11 @@
"\n",
"> **Note**\n",
">\n",
- "> * Currently, `load_in_low_bit` supports options `'sym_int4'`, `'asym_int4'`, `'sym_int5'`, `'asym_int5'` or `'sym_int8'`, in which 'sym' and 'asym' differentiate between symmetric and asymmetric quantization. Option `'nf4'` is also supported, ref... | may exclude nf3 for the time being. i.e. Keep `Option `'nf4'` is also supported, referring to 4-bit NormalFloat`.
Floating point precision -> floating point precision__s__
and mixed precisions including mixed_fp4 and mixed_fp8 are ... |
ipex-llm-tutorial | github_2023 | others | 56 | intel | hkvision | @@ -105,10 +105,11 @@
"\n",
"> **Note**\n",
">\n",
- "> * Currently, `load_in_low_bit` supports options `'sym_int4'`, `'asym_int4'`, `'sym_int5'`, `'asym_int5'` or `'sym_int8'`, in which 'sym' and 'asym' differentiate between symmetric and asymmetric quantization. Option `'nf4'` is also supported, ref... | modify as above |
ipex-llm-tutorial | github_2023 | others | 56 | intel | hkvision | @@ -17,6 +17,13 @@ pip install bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
> **Note**
> The above command will install `intel_extension_for_pytorch==2.0.110+xpu` as default
+It is also necessary to set OneAPI environment variables for BigDL-LLM on Intel GPUs. | OneAPI -> oneAPI
necessary -> required |
ipex-llm-tutorial | github_2023 | others | 56 | intel | hkvision | @@ -17,6 +17,13 @@ pip install bigdl-llm[xpu] -f https://developer.intel.com/ipex-whl-stable-xpu
> **Note**
> The above command will install `intel_extension_for_pytorch==2.0.110+xpu` as default
+It is also necessary to set OneAPI environment variables for BigDL-LLM on Intel GPUs.
+
+```bash
+# configure OneAPI env... | One -> one, same for other places |
ipex-llm-tutorial | github_2023 | others | 56 | intel | hkvision | @@ -78,7 +85,7 @@ model_in_8bit_gpu = model_in_8bit.to('xpu')
```
> **Note**
-> * Currently, BigDL-LLM on Intel GPUs has supports options `'sym_int4'`, `'asym_int4'`, `'sym_int5'`, `'asym_int5'` or `'sym_int8'`, in which 'sym' and 'asym' differentiate between symmetric and asymmetric quantization. Option `'nf4'` is... | same as above |
ipex-llm-tutorial | github_2023 | others | 56 | intel | plusbang | @@ -142,18 +130,18 @@
"\n",
"Once you have optimized the Whisper model using BigDL-LLM with INT4 optimization and loaded the Whisper processor, you are ready to begin transcribing the audio through model inference.\n",
"\n",
- "Let's start with the English audio file `audio_en.mp3`. Before we feed it ... | I feel it maybe better to provide such `wget -O audio_en.mp3 https://datasets-server.huggingface.co/assets/common_voice/--/en/train/5/audio/audio.mp3` command to make how to get `audio_en.mp3` more clear. |
ipex-llm-tutorial | github_2023 | others | 51 | intel | plusbang | @@ -0,0 +1,295 @@
+# 7.1 使用 QLoRA 微调 Llama 2 (7B)
+
+为了帮助您更好地理解 QLoRA 微调过程，在本教程中，我们提供了一个实用指南，利用 BigDL-LLM 将大语言模型对特定的下游任务进行微调。 这里使用 [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 作为示例来完成文本生成任务。
+
+## 7.1.1 在 Intel GPU 上启用 BigDL-LLM
+
+### 7.1.1.1 安装 BigDL-LLM
+
+按照[Readme](./README.md#70-environment-se... | it maybe better to use `pip install --pre --upgrade` |
ipex-llm-tutorial | github_2023 | others | 51 | intel | plusbang | @@ -0,0 +1,295 @@
+# 7.1 使用 QLoRA 微调 Llama 2 (7B)
+
+为了帮助您更好地理解 QLoRA 微调过程，在本教程中，我们提供了一个实用指南，利用 BigDL-LLM 将大语言模型对特定的下游任务进行微调。 这里使用 [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 作为示例来完成文本生成任务。
+
+## 7.1.1 在 Intel GPU 上启用 BigDL-LLM
+
+### 7.1.1.1 安装 BigDL-LLM
+
+按照[Readme](./README.md#70-environment-se... | add space to align |
ipex-llm-tutorial | github_2023 | others | 51 | intel | plusbang | @@ -0,0 +1,295 @@
+# 7.1 使用 QLoRA 微调 Llama 2 (7B)
+
+为了帮助您更好地理解 QLoRA 微调过程，在本教程中，我们提供了一个实用指南，利用 BigDL-LLM 将大语言模型对特定的下游任务进行微调。 这里使用 [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 作为示例来完成文本生成任务。
+
+## 7.1.1 在 Intel GPU 上启用 BigDL-LLM
+
+### 7.1.1.1 安装 BigDL-LLM
+
+按照[Readme](./README.md#70-environment-se... | It maybe better to comment this line and add comment "# #可以进一步减少内存使用但速度较慢" |
ipex-llm-tutorial | github_2023 | others | 51 | intel | plusbang | @@ -0,0 +1,295 @@
+# 7.1 使用 QLoRA 微调 Llama 2 (7B)
+
+为了帮助您更好地理解 QLoRA 微调过程，在本教程中，我们提供了一个实用指南，利用 BigDL-LLM 将大语言模型对特定的下游任务进行微调。 这里使用 [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 作为示例来完成文本生成任务。
+
+## 7.1.1 在 Intel GPU 上启用 BigDL-LLM
+
+### 7.1.1.1 安装 BigDL-LLM
+
+按照[Readme](./README.md#70-environment-se... | The logging information `'learning_rate': 0.0` seems a little strange to me. |
ipex-llm-tutorial | github_2023 | others | 51 | intel | plusbang | @@ -0,0 +1,295 @@
+# 7.1 使用 QLoRA 微调 Llama 2 (7B)
+
+为了帮助您更好地理解 QLoRA 微调过程，在本教程中，我们提供了一个实用指南，利用 BigDL-LLM 将大语言模型对特定的下游任务进行微调。 这里使用 [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 作为示例来完成文本生成任务。
+
+## 7.1.1 在 Intel GPU 上启用 BigDL-LLM
+
+### 7.1.1.1 安装 BigDL-LLM
+
+按照[Readme](./README.md#70-environment-se... | add indentation |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -1 +1,60 @@
-TO be added.
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning(PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLMs(large language models) using QLora with 4b... | Please kindly add a space before (
Same for other similar places :) |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -1 +1,60 @@
-TO be added.
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning(PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLMs(large language models) using QLora with 4b... | Currently, BigDL-LLM only supports QLoRA finetuning on any [Hugging Face `transformers` models](https://huggingface.co/docs/transformers/index). |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -1 +1,60 @@
-TO be added.
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning(PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLMs(large language models) using QLora with 4b... | Add "Intel Data Center GPU Max Series" |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -1 +1,60 @@
-TO be added.
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning(PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLMs(large language models) using QLora with 4b... | You could use an default conda env name here so that users could copy the command directly. |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | We may need to mention QLoRA in the introduction here. |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | of QLoRA Finetuning? |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Some dependencies are missed here. Please refer to [here](https://github.com/intel-analytics/BigDL/tree/main/python/llm/example/GPU/QLoRA-FineTuning#1-install) |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Is there any reason to specify to this specific commit? |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | I feel like it is not clear here why we need tokenizer for QLoRA finetuning :) |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | ```
Instead of `from peft import PeftModel`, we `import PeftModel from bigdl.llm.transformers.qlora` as a BigDL-LLM compatible model.
``` |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | why changing to a new line here? |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Do we need explicitly emphasize here `use_cache=True`? |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | We could directly show the difference between the results of fine-tuned model and the pretrained model. No need to show the inference code again for the pretrained model. It may confuse the reader :) |
ipex-llm-tutorial | github_2023 | others | 50 | intel | plusbang | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Please update comments of these parameters, referring to https://huggingface.co/docs/peft/conceptual_guides/lora#common-lora-parameters-in-peft |
ipex-llm-tutorial | github_2023 | others | 50 | intel | plusbang | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Should we hide the specific value of `train_runtime`, `train_samples_per_second` and `train_steps_per_second` |
ipex-llm-tutorial | github_2023 | others | 50 | intel | plusbang | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | I feel that the final LoRA weights and configurations are saved to `${output_dir}/ adapter_model.bin` and `${output_dir}/adapter_config.json`. Please confirm it. |
ipex-llm-tutorial | github_2023 | others | 50 | intel | plusbang | @@ -0,0 +1,321 @@
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the finetuning process, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is used as an exa... | Please check it as above mentioned. |
ipex-llm-tutorial | github_2023 | others | 50 | intel | plusbang | @@ -1 +1,60 @@
-TO be added.
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning(PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLMs(large language models) using QLora with 4b... | We don't need to set the two parameters during finetuning. |
ipex-llm-tutorial | github_2023 | others | 50 | intel | yangw1234 | @@ -0,0 +1,313 @@
+
+# 7.1 Finetuning Llama 2 (7B) using QLoRA
+
+To help you better understand the process of QLoRA Finetuning, in this tutorial, we provide a practical guide leveraging BigDL-LLM to tune a large language model to a specific task. [Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) is use... | import `intel_extension_for_pytorch` for using xpu device? |
ipex-llm-tutorial | github_2023 | others | 50 | intel | Oscilloscope98 | @@ -1 +1,68 @@
-TO be added.
+
+## Chapter 7 Finetune
+
+As one of the advanced parameter-efficient fine-tuning (PEFT) techniques, QLoRA enables light-weight infusion of specialty knowledge into a large language model with minimal overhead. BigDL-LLM also supports finetuning LLM (large language models) using QLora with... | You need to set OneAPI environment variables for BigDL-LLM on Intel GPUs. |
ipex-llm-tutorial | github_2023 | others | 43 | intel | shane-huang | @@ -1,5 +1,5 @@
# 第三章 应用开发基础
-本章提供了 BigDL-LLM 的基本用法以进行基础的大语言模型应用开发。
+本笔记本将帮助你开始使用 BigDL-LLM 构建一个基本的聊天应用程序。
-在 [3_Quick_Start.ipynb](./3_Quick_Start.ipynb) 中，您将学习如何一行代码加速任意 Pytorch 大语言模型并实现最基础的对话应用。 | why change this? do not change contents of chapter 3. Change from 4-8 |
ipex-llm-tutorial | github_2023 | others | 43 | intel | shane-huang | @@ -1,4 +1,4 @@
-# 第二章 环境准备
+# 第二章 环境设置 | change back to 环境准备 |
ipex-llm-tutorial | github_2023 | others | 40 | intel | shane-huang | @@ -18,34 +18,43 @@
"metadata": {},
"outputs": [],
"source": [
- "!pip install bigdl-llm[all]"
+ "!pip install --pre --upgrade bigdl-llm[all]"
]
},
{
"attachments": {},
"cell_type": "markdown",
"metadata": {},
"source": [
- "This one-line command will install `bigdl-llm` w... | `tokenizer = LlamaTokenizer.from_pretrained(model_path)` put tokenizer initialization to Run LLM section |
ipex-llm-tutorial | github_2023 | others | 40 | intel | shane-huang | @@ -71,23 +80,57 @@
"metadata": {},
"outputs": [],
"source": [
+ "save_directory='./open-llama-3b-v2-bigdl-llm-INT4'\n",
+ "\n",
+ "model.save_low_bit(save_directory)"
+ ] | also show load usage. Refer to Notebook 4.1 save/load section. |
ipex-llm-tutorial | github_2023 | others | 33 | intel | Oscilloscope98 | @@ -0,0 +1,54 @@
+# 第一章 简介
+
+## 什么是 BigDL-LLM
+[BigDL-LLM](https://github.com/intel-analytics/BigDL/tree/main/python/llm) 是一个库，可使 LLM（大型语言模型）在低成本 PC 上快速[^1]运行（无需独立显卡）。它作为开源项目 [BigDL](https://github.com/intel-analytics/bigdl) 的一部分发布，采用 Apache 2.0 许可。
+
+
+## 您能用 BigDL-LLM 做些什么
+您可以使用 BigDL-LLM 运行 _任何 [HuggingFace trans... | Please change the link accordingly. Refer to https://github.com/intel-analytics/bigdl-llm-tutorial/pull/34/files
Also change 7B -> 6B here in title |
ipex-llm-tutorial | github_2023 | others | 25 | intel | Oscilloscope98 | @@ -167,7 +167,13 @@
"source": [
"### 6.1.4.3 Stream Chat\n",
"\n",
- "ChatGLM2-6B support streaming output function `stream_chat`, which enable the model to provide a streaming response word by word. However, other models may not provide similar APIs, if you want to implement general streaming output ... | Please remove this sentence :) "The [Hugging Face *transformers* streamer classes](https://huggingface.co/docs/transformers/main/generation_strategies#streaming) is currently being developed and is subject to future changes.\n""
ChatGLM is not using streamer for stream chat
|
ipex-llm-tutorial | github_2023 | others | 17 | intel | Oscilloscope98 | @@ -0,0 +1,664 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5: Langchain Integrations \n",
+ "\n",
+ "[LangChain](https://python.langchain.com/docs/get_started/introduction.html) is a popular library for developing applications powered by language model... | It would be better to add the model link here for vicuna-7b-v1.5 :) |
ipex-llm-tutorial | github_2023 | others | 17 | intel | Oscilloscope98 | @@ -0,0 +1,664 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5: Langchain Integrations \n",
+ "\n",
+ "[LangChain](https://python.langchain.com/docs/get_started/introduction.html) is a popular library for developing applications powered by language model... | If we need to define the `langchain` version here, should we specify the package version for any other required packages for all chaptersl? |
ipex-llm-tutorial | github_2023 | others | 17 | intel | Oscilloscope98 | @@ -0,0 +1,664 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5: Langchain Integrations \n",
+ "\n",
+ "[LangChain](https://python.langchain.com/docs/get_started/introduction.html) is a popular library for developing applications powered by language model... | There would be problem for installing `chromadb` on Windows
This package will require them to install Microsoft C++ Build Tool 14.0 or greater, which could be too heavy for the tutorial. https://github.com/chroma-core/chroma/issues/189
What should we do?
|
ipex-llm-tutorial | github_2023 | others | 6 | intel | shane-huang | @@ -0,0 +1,198 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5.2: Baichuan-13B"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Overview\n",
+ "\n",
+ "This is an example shows how to run [Baichuan-13B](https://... | I think Baichuan-13B cannot run on 8G memory? You can add a reference here to chapter 2 for recommended environment setup. |
ipex-llm-tutorial | github_2023 | others | 6 | intel | shane-huang | @@ -0,0 +1,198 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5.2: Baichuan-13B"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Overview\n",
+ "\n",
+ "This is an example shows how to run [Baichuan-13B](https://... | add some descripts about why we need save and load. |
ipex-llm-tutorial | github_2023 | others | 6 | intel | shane-huang | @@ -0,0 +1,198 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5.2: Baichuan-13B"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Overview\n",
+ "\n",
+ "This is an example shows how to run [Baichuan-13B](https://... | no need to print prompt, just output is enough to show Q & A format. Change it for all outputs. |
ipex-llm-tutorial | github_2023 | others | 6 | intel | shane-huang | @@ -0,0 +1,198 @@
+{
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "# Notebook 5.2: Baichuan-13B"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Overview\n",
+ "\n",
+ "This is an example shows how to run [Baichuan-13B](https://... | remove the inference time. We'd better not show the actual inference time. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.