Questions tagged [huggingface]
16 questions
5
votes
2 answers
Are the model implementations in Hugging Face’s transformers library created by the original model authors or by Hugging Face?
I've been exploring the implementation of models like Llama in Hugging Face’s transformers library, for example:
Hugging Face's Llama model implementation.
I’m curious about how these implementations work:
Are the model codes in Hugging Face’s…
mlibre
- 175
- 4
1
vote
0 answers
Clear instructions on how to convert a model from huggingface to gguf format. The internet will love it!
Since many hours Im not able to convert the pytorch_model-00001-of-00003.bin to GGUF.
I want to convert this LLM https://huggingface.co/LeoLM/leo-hessianai-13b-chat/tree/main
Please help me. All the files are on the system but not Tutorial worked…
1
vote
1 answer
Why is DeepSeek's inference so slow?
Why is DeepSeek's inference on HuggingFace so slow (compared to Qwen, Llama, etc.)?
Geremia
- 555
- 1
- 5
- 12
1
vote
0 answers
Understanding the embeddings model (dunzhang/stella_en_400M_v5) by Alibaba. The details about the retrieve task and the s2s task
The model I am talking about is hosted here:
From the documentation:
We simplify usage of prompts, providing two prompts for most general tasks, one is for s2p, another one is for s2s.Prompt of s2p task(e.g. retrieve task): ..., Prompt of s2s…
figs_and_nuts
- 111
- 2
1
vote
2 answers
Is it possible to use LLMs by manually picking the tokens?
Suppose that we give a prompt to an LLM model, such as "what is a banana?", the LLM would start writing spitting tokens out, out of a space of tokens, until it manages to complete a textual output that resembles an answer to the question, right?
My…
caveman
- 163
- 7
0
votes
0 answers
why aren't my Training and validation losses do not reduce when fine-tuning ViTPose from huggingface
I am trying to fine-tune a transformer/encoder based pose estimation model available here at: https://huggingface.co/docs/transformers/en/model_doc/vitpose
When passing "labels" attribute to the forward pass of the model, the model returns "Training…
Soham Bhaumik
- 101
0
votes
0 answers
Why would the tokenizer for encoder-decoder model for machine translation use bos_token_id == eos_token_id? How does it know when a sequence ends?
I see on this PyTorch model Helsinki-NLP/opus-mt-fr-en (HuggingFace), which is an encoder-decoder model for machine translation:
"bos_token_id": 0,
"eos_token_id": 0,
in its config.json.
Why set bos_token_id == eos_token_id? How does it know…
Franck Dernoncourt
- 3,473
- 2
- 21
- 39
0
votes
0 answers
Formatting a numbered list into a cohesive prose paragraph using Hugging Face Inference API
I am playing with Hugging Face Inference API and I am trying to convert a numbered list into a cohesive prose paragraph. I have tried multiple models but I am not able to get things working.
I have tried GPT-2, BLOOM and T5 models, but in each case…
Sandeep
- 101
- 1
0
votes
0 answers
Quick way to extract a shopping list from audio
I want to write code to extract a list of parts from an audio file. I'm not sure how to proceed, and I would like to hear if people have any suggestions.
The audio file contains a voice saying e.g. "We should get four 16'' wheels, each with a tyre,…
Ferdinando Randisi
- 101
- 1
0
votes
0 answers
SpeechT5 Huggingface voice conversion: how to process whole input
I have followed the voice conversion example in the Huggingface blog post, and can replicate that example in a Colab session. As mentioned in the blog, the voice conversion stops consistently in the middle of the example input sentence, due to it…
G. Debailly
- 101
- 1
0
votes
0 answers
How to use llama-cpp-python to manually pick the next tokens?
When I use llama-cli, I ask models questions and they generate tokens. I see the tokens appear as they model generates them. The model randomly selects the tokens based on the random seed.
But what I want to do is to see a list of tokens, and then I…
caveman
- 163
- 7
0
votes
0 answers
Summary Generation
I want to create a summary from a list of some bullet points and keywords . Most NLP and Transformer based models are not very well suited for short sentences and bullet point.
Bullet points are usually very small 1-4 words .
Most summarisers just…
A B
- 1
0
votes
1 answer
Train my own LLM on a smaller corpus of text?
Would it be possible to train my own LLM on a smaller corpus of text, lets say some coding documentation that I then want to ask questions about using the model?
If so, are there any recommended ways of doing this, i.e is there a prebuilt…
Dylan Dijk
- 101
- 1
0
votes
1 answer
Why are some of the weights not initialized from the pretrained model checkpoint (from hugging face)?
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
# Load tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("Maltehb/danish-bert-botxo")
model =…
Sebastian Nielsen
- 401
- 1
- 4
- 11
0
votes
1 answer
My small BERT can't even overfit on a sentiment analysis task
I'm trying to train (from scratch) a miniature BERT model on SST2, a simple binary sentiment analysis task with inputs of maybe 5-20 words at a time. As you can see in my code, my approach is a little non-standard in a few ways:
The model is quite…
Jack M
- 302
- 2
- 9