Fine-Tuning Workspace

Monitor supervised fine-tuning progress, adjust run configuration, and compare checkpoint outputs.

Current Training Run
supervised-finetune-v4In Progress
llama-3-70b-instruct-v1
Step

12,402 / 36,000

Loss

0.4281

Elapsed

04:12:44

ETA

08:45:12

Dataset Configuration
hf-instruct-hq-v2 / HuggingFace

42,402 instruction samples loaded.

Epochs: 3Batch: 8Warmup: 500
Training Progress
Training and evaluation loss by checkpoint.
Evaluation Sample
Checkpoint: 12k

Prompt

Write a Python script that calculates the Fibonacci sequence up to N terms using a generator, and explain why a generator is memory efficient.

Base Model Output

Stochastic
def fib(n):
    a, b = 0, 1
    res = []
    for _ in range(n):
        res.append(a)
        a, b = b, a + b
    return res

Hallucination: failed to use a generator as requested.

Fine-Tuned Checkpoint

Aligned
def fib_gen(n):
    a, b = 0, 1
    for _ in range(n):
        yield a
        a, b = b, a + b

Success: correct implementation and explanation.

Tokens/Sec

1,240.5

Eval Accuracy

94.2%

Next Checkpoint

16k steps